Share

To design a sustainable and high-performing knowledge management (KM) program, organizations must recognize KM as the interrelationship between people, process, technology, and social capital. As much as AI suggests magic, engineering remains a key element for the success of KM and AI. KM is an organizational capability, and its process components require people to think through approaches and tactics, to remediate exceptions and recognize when a process needs improvement or no longer serves a purpose.

I have always perceived process as the poor stepchild of KM. As much as we discuss community and culture, we often relegate process and workflow to separate channels, conferences and conversations. I think process should always be referred to as process knowledge—the embodiment of the how part of knowledge, which is often, in my experience in manufacturing, the majority of the knowledge that must be developed and nurtured.

KM embraces a number of processes itself. Stan Garfield’s 16 Process Components of Knowledge Management (Knowledge Nuggets) provide an exceptional baseline for structuring these activities. However, by synthesizing his work with other core KM methodologies, such as Bergeron’s 8-step KM Lifecycle (Essentials of Knowledge Management), Bukowitz & Williams’ Tactical-Strategic Framework (The Knowledge Management Fieldbook), and Nonaka & Takeuchi’s SECI spiral (The Knowledge Creating Company), we can build an extended, end-to-end framework.

Artificial intelligence, then, belongs inside the ecosystem; it does not replace it. AI can help people retrieve, classify, summarize, translate, compare, and draft from large bodies of content. It can also scale stale material, expose overshared information, erase context, or generate a plausible statement that no source supports. NIST calls this last risk confabulation and warns that fabricated logic or citations can increase misplaced trust (NIST, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, 2024).

Beyond process in service to other KM efforts, it is crucial to apply these disciplines to process itself. These are the same concepts I’m writing about elsewhere when it comes to agent-driven processes, because agents are just another actor in a process. They require instruction. They require determined sequences, and they require a definition of what good looks like.

The practical question is therefore not, ‘Where can we add a chatbot?’ It is, ‘At which KM step can AI reduce friction while preserving source provenance, access controls, human accountability, and the ability to correct or retire knowledge?’ The following framework treats AI as a supporting mechanism inside the KM process.

This post provides a comprehensive overview of the extended KM process ecosystem, detailing how each process operates.

The Engine of Enterprise Intelligence: An Overview of the Extended KM Process Ecosystem

Managing knowledge as a strategic resource requires a coordinated set of processes. The eight lifecycle phases below extend Garfield’s core list with established KM methods and explicit controls for AI-assisted work.

A Working Rule for AI-Enabled KM

A language model is not a knowledge base. For knowledge-intensive questions, retrieval-augmented generation (RAG) combines a language model with external, inspectable sources (Lewis et al., 2020). That architecture makes sources easier to update and inspect, but it does not make every answer correct. A responsible KM implementation applies five operating rules:

• Ground the output. Use approved, current repositories and expose the passages or documents used to formulate an answer.

• Preserve permissions. Retrieval must honor the user’s access rights; AI should not become a shortcut around information governance.

• Keep provenance. Record the source, owner, version, review date, and transformation history for AI-assisted knowledge assets.

• Make the status visible. Label AI output as a draft, recommendation, or answer - not as an approved record - until an accountable person validates it.

• Evaluate with real work. Test retrieval quality, citation support, omissions, security behavior, and usefulness on representative tasks before broad deployment.

Phase 1: Strategic Alignment & Gap Analysis (The Foundation)

Before collecting content or deploying AI, a KM program must define the decisions, risks, capabilities, and outcomes it intends to support. Skipping this step produces expensive systems that answer many questions but solve few important problems.

Knowledge Audit & Gap Analysis

• Overview: Identify critical knowledge assets, map where they reside, assign ownership, and uncover gaps that create operational bottlenecks or risk.

• Strategic purpose: Separate mission-critical knowledge from accumulated content, then address gaps through training, hiring, process redesign, partnerships, or targeted acquisition.

• Where AI can assist: Entity extraction, document clustering, topic comparison, and metadata analysis can produce candidate maps of duplicated, missing, or aging knowledge. These are investigative leads, not conclusions about employee competence or strategic value.

Social Network Analysis (SNA)

• Overview: Map the informal relationships and trust pathways through which information actually flows, rather than relying only on the formal organization chart.

• Strategic purpose: Identify knowledge brokers, isolated groups, and potential bottlenecks so the KM team can strengthen connections without overloading a few people.

• Where AI can assist: Network analytics can surface interaction patterns in collaboration metadata. Privacy review, data minimization, consent where appropriate, and bias checks are essential; message volume is not a reliable proxy for expertise, trust, or value.

Appreciative Inquiry & Positive Deviance

• Overview: Identify people or teams that achieve exceptional outcomes under comparable constraints and study the practices behind their results.

• Strategic purpose: Harvest and scale what is already working instead of treating every KM intervention as a repair project.

• Where AI can assist: Pattern analysis can help find candidate cases and compare narratives, metrics, and artifacts. People still have to determine whether the result reflects a transferable practice, a local condition, or a misleading correlation.

Phase 2: Knowledge Creation & Elicitation

Knowledge creation is generative in the organizational sense: people develop new ideas, test them, interpret experience, and expand intellectual capital. Generative AI can contribute language and combinations, but novelty in wording is not the same as validated new knowledge.

Knowledge Creation (The SECI Spiral)

• Overview: Nonaka and Takeuchi describe knowledge creation as a continuing conversion between tacit, experience-based knowledge and explicit, codified knowledge.

• Four modes: Socialization shares experience directly; externalization expresses tacit insight; combination reorganizes explicit material; internalization turns codified knowledge into practiced capability.

• Where AI can assist: Speech-to-text, summarization, comparison, and drafting can reduce the effort required for externalization and combination. AI cannot observe all the situational judgment, embodied skill, relationships, or meaning that make tacit knowledge useful; experts must review what was lost or distorted.

Knowledge Harvesting

• Overview: Use structured interviews, observation, artifacts, and follow-up questions to elicit critical expertise before it becomes unavailable.

• Strategic purpose: Reduce knowledge-loss risk while avoiding the fiction that a transcript turns unique competence into a complete corporate asset.

• Where AI can assist: With permission, AI can transcribe interviews, draft summaries and playbooks, flag unanswered questions, and compare interviews for recurring themes. The knowledge holder and a trained interviewer should validate the resulting artifact, including exceptions and boundary conditions.

Storytelling, Narrative, & Anecdotes

• Overview: Use stories to communicate context-rich experience, turning points, tradeoffs, and lessons in a memorable form.

• Strategic purpose: Preserve the human context that procedures and data often omit, while making clear which parts are evidence, interpretation, or metaphor.

• Where AI can assist: AI can outline interview material, create audience-specific drafts, or translate a story into a case, checklist, or briefing. The original narrator and an editor should protect voice, intent, attribution, confidentiality, and factual sequence.

Phase 3: Capture, Curation & Workflow Integration

Once knowledge is generated or elicited, it must be captured, validated, and integrated into work. AI is most useful here when it converts an existing, reviewable event or artifact into a draft - not when it invents a record after the fact.

Knowledge Capture & Modification

• Overview: Document and refine knowledge through editing, access controls, source linking, versioning, and approval so that it is reliable and usable.

• Strategic purpose: Filter raw inputs before the repository becomes a digital junkyard.

• Where AI can assist: AI can summarize a resolved case, suggest a draft knowledge article, detect near-duplicates, and propose metadata. ServiceNow, for example, documents a workflow in which Now Assist generates a draft article from a case or incident and then directs the agent to review and edit it before publishing. The review step is the KM control, not an optional courtesy.

Workflow & Methodology Integration (In-the-Flow KM)

• Overview: Embed capture and reuse into routine work so that KM does not become a separate task that busy employees can ignore.

• Strategic purpose: Collect knowledge close to the moment of use, when sources and context are still available.

• Where AI can assist: Meeting, incident, project, and service workflows can trigger a transcript, summary, proposed action list, or draft article. The trigger should be transparent, scoped, secure, and connected to an owner and approval path rather than becoming ambient surveillance.

Phase 4: Organization, Packaging & Translation

A repository is useful only when people and systems can interpret its content. AI raises the value of good organization because retrieval depends on clean sources; it also raises the cost of poor organization because an authoritative-sounding answer can blend contradictory documents.

Classification, Taxonomies, & Metadata

• Overview: Define a controlled vocabulary, taxonomy, and metadata schema that describe an asset’s subject, context, owner, sensitivity, status, and validity.

• Strategic purpose: Create consistent pathways for navigation, filtering, access control, analytics, and search.

• Where AI can assist: Classifiers and language models can propose tags, entities, synonyms, relationships, and candidate taxonomy updates at scale. A taxonomy owner should approve changes, monitor drift, and use reviewer feedback to improve rules and examples.

Translation & Repurposing (Knowledge Packaging)

• Overview: Translate specialized knowledge into formats and language that other audiences can understand and act upon.

• Strategic purpose: Prevent cognitive disconnects across organizational boundaries without stripping away necessary technical, legal, or cultural nuance.

• Where AI can assist: AI can produce summaries, translations, glossaries, FAQs, role-specific explanations, and draft training material. Subject-matter review is essential whenever simplification could alter scope, safety conditions, obligations, or meaning.

Content Management

• Overview: Manage documented assets from draft through review, approval, publishing, maintenance, and retirement.

• Strategic purpose: Keep the active knowledge base current, credible, traceable, and aligned with corporate standards.

• Where AI can assist: Automated checks can flag broken links, inconsistent terms, missing owners, near-duplicates, sensitive content, and review dates. These signals prioritize editorial work; they do not establish truth or authorize publication.

Phase 5: Dissemination, Transfer & Search

This phase gets the right knowledge to the right people at the point of need through both push and pull mechanisms. A widely cited 2012 McKinsey analysis estimated that interaction workers spent 19 percent of their week searching for and gathering information. That figure is a dated baseline, not a universal current measure, but the friction it describes remains central to KM.

Knowledge Transfer & Dissemination

• Overview: Move tacit and explicit knowledge across time, distance, and organizational boundaries through coaching, communities, briefings, learning, messaging, and repositories.

• Strategic purpose: Spread solutions and know-how while preserving the context needed to apply them safely.

• Where AI can assist: Recommendation and generation systems can suggest relevant content, draft role-specific briefings, and create practice questions. People should be able to see why an item was recommended and still reach human experts, especially when the question is novel or consequential.

User Access, Search, & Expertise Location

• Overview: Use enterprise search, semantic retrieval, directories, and communities to help people find both documented knowledge and human expertise.

• Strategic purpose: Reduce search time, improve reuse, and connect seekers to people who can interpret ambiguous or tacit knowledge.

• Where AI can assist: Semantic search can retrieve conceptually related material even when the query and source use different words. RAG can then synthesize retrieved passages into a conversational answer with source links. Microsoft documents this pattern in Microsoft 365 Copilot, which grounds responses in Microsoft Graph content the signed-in user is permitted to access. Grounding improves relevance; it does not eliminate unsupported claims, so citations and an honest ‘not enough evidence’ response remain essential.

Phase 6: Application, Reuse & Continuous Learning

The test of KM is not how much content the organization stores or how many answers an AI system generates. It is whether people apply appropriate knowledge to improve performance, avoid repeat mistakes, and learn when circumstances change.

Knowledge Reuse & Proven Practices

• Overview: Adapt previous deliverables, replicate proven practices, and reuse standardized procedures where the context supports doing so.

• Strategic purpose: Reduce reinvention without turning yesterday’s solution into today’s unquestioned rule.

• Where AI can assist: AI can retrieve analogous cases, compare conditions, propose a starting template, and highlight differences. The accountable practitioner decides whether the precedent applies and documents any adaptation.

Lessons Learned

• Overview: Use after-action reviews and retrospectives to capture what happened, why participants believe it happened, and what future teams should test or change.

• Strategic purpose: Institutionalize organizational memory while treating lessons as context-dependent hypotheses, not timeless facts.

• Where AI can assist: AI can transcribe a review, group recurring themes, retrieve similar lessons, and draft actions. It should preserve disagreement and uncertainty rather than manufacture consensus or causal certainty.

A useful empirical example comes from customer support. Brynjolfsson, Li, and Raymond studied an AI assistant used by more than 5,000 agents and reported a 15 percent average productivity increase, with larger gains for novice and lower-skill workers (2025). The evidence is consistent with AI helping disseminate patterns embodied in previous interactions. It does not prove that the same result will appear in every KM process, workforce, or task.

Phase 7: Evaluation, Metrics, & Change Management

KM and AI both require continuous evaluation. A pilot that impresses in a demonstration can still fail on access control, rare queries, changing content, or the tasks that matter most.

KM Metrics, Reporting, & Valuation

• Overview: Track KM performance through goal, operational, learning, risk, and business-impact measures.

• Strategic purpose: Show whether the program changes decisions, cycle time, quality, resilience, learning, or cost - not merely whether people opened a portal.

• Where AI changes the metrics: Add retrieval precision, source coverage, citation relevance, groundedness, correction rates, sensitive-data exposure, answer abstention, freshness, latency, and human escalation. Measure time saved and outcome quality separately; faster wrong answers are not productivity.

Change Management

• Overview: Manage the cultural and behavioral shifts required for open sharing, responsible reuse, and appropriate trust in AI-assisted work.

• Strategic purpose: Create adoption without confusing compliance, speed, or novelty with learning.

• Where AI changes the work: Build AI literacy, role-specific guidance, feedback channels, escalation paths, and time for experts to curate what the system uses. Research with 758 consultants found strong gains on tasks inside the model’s capability frontier but emphasized that performance varies by task - the ‘jagged technological frontier’ (Dell’Acqua et al., 2026). KM governance must therefore be task-specific.

Phase 8: Archiving, Disposal & Divestiture

Knowledge becomes stale when it is not maintained. AI-assisted retrieval makes lifecycle management more important, not less: obsolete content can be rediscovered and fluently restated long after people have stopped opening the original file.

Archiving & Maintenance

• Overview: Review, update, archive, and secure older explicit content while maintaining its provenance and retention status.

• Strategic purpose: Keep active search paths clear and reduce the chance that people or AI systems rely on outdated, conflicting sources.

• Where AI can assist: AI and rules can flag low-use items, expired review dates, version conflicts, broken links, and material that contradicts a newer approved source. An owner should decide whether to update, supersede, restrict, or archive it.

Disposal & Divestiture (Forbearance)

• Overview: Purposefully destroy, prune, license, sell, or decline to maintain knowledge assets that are obsolete, risky, duplicative, or no longer strategically aligned.

• Strategic purpose: Remove zombie knowledge while meeting legal holds, records schedules, intellectual-property obligations, and evidence requirements.

• Where AI can assist: AI can prioritize candidates and assemble evidence for review. It should never be the sole authority for deletion, divestiture, or records disposition; those actions require accountable owners and applicable legal, security, and records-management approval.

Treat Agents as a KM Lifecycle Participants

An agent should not sit outside the KM process. It is another actor with delegated access to knowledge, tools, and decisions. The workflow, instructions, examples, permissions, memory, and escalation rules that guide it all become process knowledge—and must be managed accordingly.

KM disciplines should therefore govern the agent’s entire operating environment:

• Align: Define the outcome, scope, constraints, and accountable owner.

• Ground: Limit retrieval to approved, current, permission-appropriate sources.

• Capture: Record inputs, outputs, sources, transformations, decisions, and exceptions.

• Curate: Treat instructions and agent-created artifacts as versioned assets with owners and review dates.

• Evaluate: Test performance against real cases, including ambiguity, conflicting sources, security boundaries, and failure conditions.

• Escalate and retire: Specify when human judgment is required and remove obsolete instructions, tools, sources, and agent-generated knowledge.

The goal is not to make agents understandable, correctable, and governable. If a traditional process requires ownership, evidence, review, and lifecycle management, its agent-enabled counterpart requires the same controls, usually more explicitly.

Summary Matrix: Extended Process Lifecycle

Conclusion: AI Amplifies the Process It Enters

AI can make KM more conversational, responsive, multilingual, and usable in the flow of work. It can lower the cost of turning a case into a draft article, an interview into a playbook, or a question into a source-linked answer. Those are meaningful improvements.

But AI does not rescue an organization from missing ownership, weak permissions, stale content, absent review, or a culture that withholds context. It amplifies the process it enters. Put it inside a disciplined KM lifecycle and it can expand access to organizational knowledge. Put it on top of a digital junkyard and it will make the junk easier to retrieve, remix, and trust.

Subscribe now

References

National Institute of Standards and Technology. (2024). Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST AI 600-1). DOI and full report.

Lewis, P., et al. (2020). Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. NeurIPS 2020. Paper.

Brynjolfsson, E., Li, D., & Raymond, L. R. (2025). Generative AI at Work. The Quarterly Journal of Economics, 140(2), 889-942. DOI.

Dell’Acqua, F., et al. (2026). Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality. Organization Science, 37(2), 403-423. DOI.

McKinsey Global Institute. (2012). The Social Economy: Unlocking Value and Productivity Through Social Technologies. Full report. The report’s 19 percent figure describes searching and gathering information, not a current universal measure of time lost.

Microsoft. (updated 2026). Microsoft 365 Copilot architecture and how it works. Microsoft Learn.

ServiceNow. (updated 2026). Generate a Knowledge article from the Now Assist panel. Product documentation. This is a documented product workflow, not independent evidence of effectiveness.

Bergeron, B. (2003). Essentials of Knowledge Management. Wiley.

Bukowitz, W. R., & Williams, R. L. (1999). The Knowledge Management Fieldbook. Financial Times/Prentice Hall.

Nonaka, I., & Takeuchi, H. (1995). The Knowledge-Creating Company. Oxford University Press.

When organizations embark on a knowledge management journey, their first instinct is often to buy something. A new platform. A better search engine. An AI assistant. A repository into which the organization can finally pour everything it knows.

This is understandable. Technology is visible, purchasable, and relatively easy to put on a project plan. Policy, practice and how we use space are none of those things. Some call it culture; I try to avoid that term and focus on the levers that enable it, and empower people to behave positively, whatever positive means to the target organization.

But knowledge management is not primarily a system installation. It is an attempt to change how people seek help, share experience, make decisions, learn from mistakes, and preserve what matters to people and to the organization.

Technology can reduce friction and lower barriers to capture. It cannot determine what is worth knowing, supply missing context, or create the trust required to share experience.

A repository without contributors is an empty room. Search without trust produces guarded answers. An expertise directory is useless if asking for help is interpreted as weakness. Generative AI can synthesize whatever an organization has captured, but it cannot recover the experience people were never given the time or safety to share.

The real unit of knowledge management is not the document. It is the human interaction that creates, tests, transfers, and applies knowledge.

That makes KM a people system supported by technology.

Subscribe now

The eleven people components

’s KM framework identifies eleven people components that organizations should intentionally design and manage. Together, they provide a useful baseline for the human side of a KM program.

1. Knowledge-sharing culture and values

Culture is the accumulated answer to a simple question: What happens here when someone shares what they know? I see this as encapsulated in policy, practice, space, and yes, technology- not any technology, but specific technology aimed at that question.

Are people thanked or ignored? Does admitting uncertainty invite help or judgment? Are lessons from failure used to improve the system or identify someone to blame?

Posters about collaboration cannot overcome incentives, management practices, and promotion decisions that reward individual knowledge hoarding. A knowledge-sharing culture emerges when everyday behavior demonstrates that asking, contributing, reusing, and learning are valued.

2. Knowledge managers and KM leaders

KM requires people who can translate organizational goals into knowledge flows.

Their work includes setting strategy, building governance, supporting communities, facilitating knowledge capture, improving findability, measuring outcomes, and connecting KM activity to business performance.

This is not repository administration with a more fashionable title. A capable knowledge manager works across organizational boundaries and understands enough about people, process, content, technology, and change to not just hold the system together, but to inspire its use.

3. User and employee satisfaction surveys

KM teams frequently measure what their platforms can count: page views, downloads, searches, posts, and active users.

Those measures describe activity and often fail to capture value.

Surveys and interviews can reveal whether people are finding answers faster, avoiding repeated work, making better decisions, and experiencing less frustration. They can also expose what dashboards miss: distrust of the content, difficulty knowing what is current, reluctance to ask questions, or the belief that contributing is someone else’s job.

The most useful survey question may be the simplest: What gets in the way of doing your work?

4. Social networks

The organization chart shows reporting relationships. It does not show how knowledge actually moves.

Employees rely on informal networks of trusted colleagues: the person who remembers why a decision was made, the veteran who knows which procedure does not work in practice, or the connector who can locate an expert in another part of the organization.

KM should make these networks easier to navigate without trying to bureaucratize every relationship. Social and organizational network analysis can help identify connectors, isolated teams, overburdened experts, and dangerous single points of dependency.

5. Communities of practice

Communities of practice bring together people who share a professional domain and want to improve their practice.

A good community is not merely a distribution list or a monthly presentation. It is a place where members can compare experience, solve problems, develop shared approaches, and gradually expand the organization’s collective capability.

Communities need a clear domain, an active core group, useful activity, and enough autonomy to respond to members’ needs. They also require time. A community cannot thrive indefinitely on volunteer labor squeezed into the margins of the working day.

See for more:

6. KM training

KM training should teach more than which button to press to assign metadata.

Employees need to know how to formulate a useful question, identify reusable knowledge, distinguish evidence from opinion, document context, assess the credibility of a source, and adapt someone else’s experience to a new situation.

In an AI-enabled workplace, these capabilities become even more important, not less. Employees must be able to ask questions with intent, evaluate generated answers, inspect sources, recognize uncertainty, and know when human expertise is required.

7. KM documentation

People need concise guidance explaining how the KM environment works.

This includes overviews, contribution standards, governance policies, role descriptions, templates, and practical “how-to” material. Documentation should answer the questions users actually have, in language they recognize, at the moment they need help.

The irony of unreadable KM documentation should not be lost on us.

In the world of AI, this is a perfect starting point for a partnership. The KM team can reimagine their content for AI ingestion. They will find this is not a trivial activity. That experience can then be applied to other projects as they help teams ready their knowledge content for easier and more accurate generative AI retrieval.

8. KM communications

KM must be continually explained through examples of value.

A successful communication program does more than announce features. It tells stories: how one team reused another team’s work, how an expert connection prevented a mistake, how a community shortened the learning curve, or how a captured lesson changed a decision.

These stories make an abstract capability concrete. They also help employees see KM as part of the work rather than another corporate program competing with it. And they help those holding the budget strings think twice, especially when the story includes real numbers for savings or new revenue.

9. User assistance and the knowledge help desk

Search is not always enough.

Sometimes the seeker does not know the correct terminology. Sometimes the knowledge is fragmented. Sometimes the answer exists only in someone’s experience. A knowledge help desk, “ask the expert” service, or research function can turn a dead end into a useful connection.

The goal should not be to answer every question centrally. It should be to ensure that questions reach the right content, community, or person—and that recurring questions improve the knowledge environment.

Again, this is a great place to start with AI. Learn how to teach AI to answer these questions effectively. That doesn’t mean the help desk gets retired, just that the people can focus on better answers and tackle emergent questions for which no documentation yet exists.

10. KM goals and measurements

KM goals should describe changes in organizational performance and behavior, not simply the production of content.

Useful measures might include time saved locating expertise, reduction in repeated mistakes, speed to competence for new employees, reuse of proven practices and the related increase in quality, response time to questions, or the number of decisions improved by access to prior experience.

Counting contributions is easy. Demonstrating that knowledge changed an outcome is harder—and much more important.

11. KM incentives and rewards

Recognition matters, but incentive design requires care.

If employees are rewarded for the number of documents they upload, the organization will receive more documents. It may not receive more useful knowledge.

Recognition should emphasize usefulness, reuse, responsiveness, mentoring, and contribution to collective outcomes. It can be formal, but often the most powerful reward is professional reputation: becoming known as someone who helps others succeed.

Garfield’s eleven components form a strong structural foundation. But structure alone does not fully explain who performs the work, why people participate, or how human systems remain healthy over time. Three additional dimensions complete the picture.

Subscribe now

The first missing dimension: intermediary roles

“Knowledge manager” is often used as an umbrella term for very different kinds of work.

Research associated with IBM’s early knowledge-management programs distinguished among several forms of knowledge intermediation. Three are especially useful when designing KM roles.

Knowledge stewards

Knowledge stewards capture, organize, maintain, and improve knowledge assets.

They interview experts, observe work, facilitate after-action reviews, identify reusable practices, and turn experience into forms other people can understand. They also ensure that important material has an owner, context, review date, and retirement path.

Their job is not simply to collect information. It is to preserve meaning.

Knowledge brokers

Some knowledge is too contextual, nuanced, or “sticky” to be separated from the person who holds it.

Knowledge brokers connect people rather than attempting to document everything. They understand enough about the organization to recognize who should speak with whom, make the introduction, and help frame the exchange.

Their product is not a document. It is a productive conversation.

Knowledge researchers

Knowledge researchers find and synthesize information in response to a need.

They search internal and external sources, evaluate credibility, identify emerging developments, and push relevant intelligence to decision-makers. Librarians, competitive-intelligence professionals, and information specialists often perform this role particularly well. In a corporate setting, librarianship, competitive intelligence, and information research should be explicitly connected to the organization’s KM operating model.

These three archetypes support different movements of knowledge:

• Stewards move knowledge from people into durable assets.

• Brokers move knowledge between people.

• Researchers move relevant information toward a problem or decision.

Expecting one generic “KM person” to perform all three functions equally well usually leads first to role-design failure, and then failure of the knowledge management system that poor person was expected to support.

The second missing dimension: motivation

Organizations often assume that people will share knowledge if they are instructed, reminded, or rewarded.

Human motivation is more complicated.

Self-determination theory offers a useful lens. It suggests that people are more likely to sustain a behavior when three psychological needs are supported: autonomy, competence, and relatedness. Research applying the theory to knowledge sharing similarly points toward the importance of self-directed, internalized motivation rather than mere compliance.

For KM, that means:

Autonomy: Give people meaningful choice in how they contribute. Do not turn every insight into a mandatory form with nineteen required fields.

Competence: Help employees feel capable of contributing something useful. Provide examples, coaching, templates, editorial support, and constructive feedback.

Relatedness: Make contribution part of belonging to a professional community. People share more readily when they trust the audience and believe their contribution will help someone they recognize.

This does not mean abandoning goals or recognition. It means avoiding incentives that crowd out the reason people most want to share: to solve a problem, help a colleague, improve the practice, and be valued for what they know.

A points system may produce activity. Purpose produces commitment.

The third missing dimension: dynamic community leadership

Communities of practice are often described as self-organizing. This is sometimes misinterpreted to mean self-sustaining.

They are not the same thing.

Healthy communities require leadership, but not necessarily traditional hierarchical leadership. The European Commission’s Communities of Practice Playbook, for example, emphasizes both internal community leadership and organizational sponsorship.

Several roles are usually present:

• A sponsor creates legitimacy, protects time, and removes organizational barriers.

• A community leader or coordinator connects members and maintains momentum.

• A core group shares responsibility for direction and programming.

• Subject-matter experts deepen the practice.

• Connectors bring in people and ideas from adjacent networks.

• Members move between peripheral and active participation as their needs and circumstances change.

Leadership should therefore be distributed and dynamic. The person who convenes the community does not have to provide all its expertise. The most active members today may become less active tomorrow. New leaders should be cultivated rather than treating the founding coordinator as a permanent source of energy.

A community is resilient when leadership can circulate.

Turning the components into an operating system

The temptation at this point is to turn the people ecosystem into another large implementation program.

I would start smaller.

Begin with a consequential knowledge flow

Choose a business problem where knowledge matters: onboarding, customer escalation, proposal development, equipment maintenance, product delivery, or the loss of experienced employees.

Map what currently happens.

Where does the question begin? Who is asked? What can be found? Where does the process stall? Which expert becomes a bottleneck? What is repeatedly recreated? What knowledge disappears when the work ends?

This gives the KM effort an operational purpose. Interestingly, this is the same advice I give to organizations seeking to create and deploy AI-powered agentic workflows. Agents become another tool that can shape the design, not a replacement for the documentation or the workflow.

Design the human roles

Decide who will steward critical knowledge, broker connections, conduct research, lead communities, sponsor the effort, and maintain the supporting environment.

Give these roles explicit time and authority. “Do this when you can” is not a role description.

Create the conditions for contribution

Make it safe to ask questions and admit uncertainty. Reduce the effort required to contribute. Provide editorial and facilitation support. Show employees where their contributions were reused.

Most importantly, ensure that managers do not publicly endorse knowledge sharing while privately rewarding only individual utilization and delivery.

Build feedback into the work

Combine system data with interviews, surveys, stories, and operational measures.

Watch what people do when they need an answer, not only what they say they do. Repeated workarounds are not user failures. They are design evidence.

Add technology where it removes friction

Once the knowledge flow, roles, and behaviors are understood, technology choices become clearer.

The organization may need better search, a community platform, an expertise locator, workflow integration, knowledge analytics, or an AI assistant. But the tool is now being selected to support an ecosystem, not being selected as the framework for defining a new ecosystem forced on users by the tool’s capabilities.

The leadership test

Leaders who want a knowledge-sharing organization should ask themselves:

• Do people have time to share and reuse knowledge?

• Is asking for help treated as responsible behavior?

• Are experts rewarded for enabling others, or only for their individual output?

• Can employees locate a person when the answer cannot be documented?

• Does important knowledge have a steward?

• Do communities have active leadership and sponsorship?

• Are lessons used for learning or blame?

• Are KM measures connected to business outcomes?

• Would employees still participate if the points, badges, and campaigns disappeared?

The answers reveal more about KM readiness than a technology inventory ever will.

The central lesson is simple: knowledge does not flow because a platform exists. It flows when people have a reason to share, a safe place to do it, a practical way to connect, and evidence that their contribution matters.

The future of knowledge management will undoubtedly include more powerful search, analytics, automation, and artificial intelligence. But these technologies will amplify the knowledge environment they are given.

If that environment is fragmented, distrustful, and poorly maintained, AI will make the fragmentation easier to query, but it will not return better answers.

If it is connected, curious, well-stewarded, and generous, technology can amplify the accumulated knowledge.

Perhaps the path is not really from bits to brains. It runs in the other direction. Knowledge begins with people—with their experience, judgment, relationships, and questions. Technology can capture and amplify some of it, but the sequence matters.

Brains before bits.

Click for SI AI Insights

This post was inspired by Thariq Shihipar’s Claude/Anthropic post: Using Claude Code: The unreasonable effectiveness of HTML.

Every few years, some technically elegant shortcut escapes its original context and becomes a proposed standard for everyone else. Markdown has followed that path. It began as a lightweight way to write for the web without constantly touching HTML. For developers, documentation teams, and people living inside text editors, it solved a real problem. It made structured text portable, readable, and easy to version.

Fine. Developers should use whatever works for them.

But that does not mean writers should accept Markdown as the future of writing. From a writer’s point of view, and from the perspective of knowledge management, Markdown is too thin a container for meaning. It confuses visible formatting conventions with structure. It encourages people to think that because a document can be rendered cleanly, it has been adequately described.

It has not.

Markdown is a useful notation. It is not a rich knowledge model.
I lived through the earlier era of dot commands, reveal codes, control characters, embedded formatting instructions, and all the other visible machinery that once sat between the writer and the work. Those systems were not romantic. They were tedious. They forced writers to think like operators. They made us accommodate the limitations of machines and software.

Markdown often feels like a return to that mindset, only with better branding.
Yes, Markdown is cleaner than old formatting codes. Yes, it is easier to read than raw HTML. But the deeper issue remains: the writer is still being asked to manually carry the structural burden. Add the hash marks. Add the asterisks. Add the brackets. Add the backticks. Learn the dialect. Remember which flavor supports tables, footnotes, task lists, callouts, embeds, and metadata. Hope the next system interprets the file the same way the last system did.

That is not progress. That is a compromise being mistaken for a destination.

Writers Need More Than Rendered Text

Writers do not just produce strings of words. Writers create arguments, examples, evidence, transitions, citations, asides, definitions, claims, counterclaims, captions, pull quotes, notes, summaries, and narrative movement. Those elements have relationships. They have purpose. They have weight. Markdown flattens too much of that work.

Markdown flattens meaning

A Markdown heading may indicate hierarchy, emphasis, navigation, a topical shift, or simply a formatting preference. A list may represent sequence, options, priorities, evidence, instructions, requirements, or loosely related ideas. A block quote may be an actual quotation, an excerpt, a design convention, or a highlighted passage. Markdown rarely knows the difference.

That distinction is important. Knowledge management has always struggled with the gap between information storage and knowledge representation. Markdown stores content efficiently, but it does not adequately capture its context. It does not naturally capture why something matters, how it relates to other elements, what role it plays in an argument, or how it should be treated by a downstream system.

HTML, used well, gets much closer.

The argument for HTML

HTML can distinguish an article from a section, a figure from a caption, a citation from a casual link, an aside from the main body, navigation from content, and metadata from prose. It can carry identifiers, classes, attributes, accessibility cues, schema, provenance, and machine-readable relationships. HTML can render beautifully for people and remain intelligible to machines.
That does not mean writers should hand-code HTML. That would be another regression.

The real argument is that authoring tools should let writers work with meaningful structures and generate semantic HTML beneath the surface. The writer should not have to manage the machinery. The machine should. That holds true even if HTML is supplanted by an even richer, more subtle representation of knowledge (see SIDEBAR: The Formats That Were Already Trying to Solve This, for background on previous attempts).

AI Makes Thin Structure More Dangerous

AI’s role in the content lifecycle makes the argument against Markdown more urgent.

A document no longer simply waits to be read. AI systems summarize it, chunk it, index it, retrieve it, cite it, remix it, analyze it, convert it into training material, ingest it into knowledge bases, and drive automated workflows with it. Most modern documents become operational.

That means structure matters more, not less.

AI systems do not need more vaguely formatted text. They need better context. They need to know what is central and what is supporting. They need to know whether a passage is evidence, an example, a caveat, a quote, an instruction, a source, a claim, or metadata. They need to distinguish the body of the argument from the sidebar. They need to preserve captions, citations, dates, authorship, versioning, and relationships among ideas.

Markdown gives AI a rough outline. Semantic HTML can give AI a map.

That difference matters for writers and for organizations. AI ingestion without rich context produces brittle results. It increases the chance that systems will treat all content as equivalent, detach statements from their evidentiary base, overlook authorial intent, or misrepresent supporting material as the main argument.
For organizations trying to manage knowledge, Markdown lacks sufficient connective tissue. It makes documents portable, but portability is not the same as meaning. It creates lightweight files, but lightness is not the same as usefulness. It supports rendering, but rendering is not the same as understanding.

SIDEBAR: The Formats That Were Already Trying to Solve This

The argument for semantic richness is not new. The technical writing and documentation communities have been making it for decades, and they built real standards to back it up.

DITA, short for the Darwin Information Typing Architecture, emerged from IBM in the early 2000s as a way to structure content at the topic level rather than the document level. DITA forces authors to classify content as a concept, a task, or a reference before they write a word. That classification shapes everything: how content is reused, how it is assembled and how downstream systems understand what they are retrieving. DITA is not easy. It is opinionated, verbose, and built for enterprise environments with dedicated toolchains. But the underlying premise, that content type is a first-class attribute, not an afterthought, is exactly right.

DocBook took a different approach. It defined a rich XML vocabulary for technical documentation, allowing authors to tag content with enough precision that a single source could render as a printed manual, an online help system, a PDF, or a web page without losing structural integrity. DocBook’s granularity was its strength and its burden. It provided the map. It also required knowing how to read one.

Neither format caught on outside of specialized technical communities, and that failure is worth understanding. Both demanded too much of authors and too little from tools. They solved the knowledge representation problem and ignored the user experience problem. Writers did not want to manage XML tags; they wanted to write. The gap between the format and the authoring environment was never adequately closed.

That gap is where AI-native knowledge formats now have an opportunity. Projects like LlamaIndex’s document schemas, emerging structured knowledge graphs, and purpose-built retrieval-augmented generation pipelines each, in their own way, attempt to rebuild what DITA and DocBook tried to do — preserve structure, intent, and relationships — but in environments where AI can assist rather than just consume. The authoring layer is still catching up to the architecture.

The lesson from DITA and DocBook is not that structured authoring failed. It is that structure without usability is a standard without an audience. The next iteration needs to get both right.

The False Romance of Plain Text

Markdown’s advocates often celebrate plain text as durable, transparent, and future-proof. There is truth in that. Plain text has advantages. It survives platform churn. It can be searched, versioned, compared, and moved.

But writers should be wary of any argument that treats plain text as inherently superior because it is simple.

Simple for whom?
Markdown is often simple for people who already think in systems, repositories, parsers, and pipelines. It is less simple for people whose work depends on shaping meaning rather than managing syntax.

A good authoring environment should not ask writers to choose between expressive richness and technical durability. Personal computing was supposed to move us past that bargain. We have decades of experience in interface design, document modeling, structured authoring, metadata management, accessibility work, and web standards. We should not be telling writers that the best future we can offer is a text file filled with appropriated punctuation.

The future of writing should not look like a cleaned-up version of the command-line past.

The future should let writers create structured, semantically rich, machine-readable work without forcing them to see or manipulate the underlying codes. Writers should be able to identify a passage as an example, quote, warning, sidebar, citation, definition, or reusable knowledge object through the authoring interface. The system should preserve that structure in durable, open, inspectable formats.

HTML isn’t perfect, but it is closer to that future than Markdown.

Markdown Still Has a Place

None of this requires declaring Markdown useless. Markdown is excellent for notes, README files, lightweight documentation, quick drafts, developer workflows, and situations where speed matters more than semantic richness. It is a good scratchpad. It is a good interchange convenience. It is often good enough.

But “good enough” should not become the default ambition for writing, publishing, and knowledge management in the AI era.

The mistake is not using Markdown. The mistake is pretending that Markdown can carry the full burden of modern content.

Serious writing needs more than headings, bullets, links, and emphasis. Serious knowledge work needs containers that preserve relationships, context, intent, provenance, accessibility, and reuse. The closer content gets to AI ingestion and organizational memory, the less acceptable it becomes to flatten meaning into decorative shorthand.

Toward A Better Standard

The better standard is not “writers should write HTML.”

The better standard is this: writers should write in tools that understand knowledge structure, and those tools should produce rich semantic HTML or similarly expressive open formats underneath.

That distinction is important. Writers should not be asked to become markup clerks. They should be supported as creators of meaning. The authoring system should make structure easy to apply, inspect, preserve, and interpret by AI systems.

Markdown was a humane compromise for a less capable computing environment. It reduced friction at a time when writing for the web was harder than it needed to be, and it helped reduce the cost of token processing. But compromises have a shelf life. When a compromise starts limiting thought, context, reuse, accessibility, and machine understanding, it stops being liberating.

Markdown helped make web writing simpler, but simple isn’t the goal of communication, knowledge capture and knowledge representation. We live in a messy world, and we have spent decades devising ways to capture the underlying richness of knowledge and experience in our digital artifacts. If we want AI to understand the world, it also needs to embrace its complexities.

Organizations get into trouble when they treat knowledge management as if it were a destination rather than a discipline. They announce a KM vision, launch a platform, form a steering committee, rename a team, and then wait for knowledge to behave as though it has a life of its own. It does not. Knowledge creates value only when it improves how the organization decides, executes, learns, and adapts.

Read more

In most organizations, KM doesn’t fail because the platform is wrong. KM fails because leadership treats “knowledge-sharing” as a value statement instead of a management practice. Enterprise KM becomes real only when the C-suite moves from approval to participation…when they start actively designing the narrative, funding the work, protecting the practice, and showing up inside the system.

1. The Architect: weaving knowledge into the company narrative

A senior leader’s first job in KM is sense-making. KM has to stop sounding like an “initiative” and start acting like a strategy. Leaders create the context that orients the organization’s attention, and that translates into what it preserves, what it reuses, and what it chooses to forget.

That’s why the commitment isn’t “encourage sharing.” It’s to learn to speak KM as part of the business story, personally, credibly, and repeatedly. Leaders who can give the KM presentation themselves without delegating the narrative to a program office forcefully demonstrate that knowledge is not a side job, extra work, or an unrewarded project. Regular communication about KM in the channels people actually notice (town halls, leadership updates, operating reviews) turns KM from background noise into a management expectation.

This also has an operational expression: KM belongs on the leadership agenda. When KM becomes a standing topic: knowledge gaps, reuse wins, community signals, emerging lessons, ties to revenue and cost savings, then the organization learns that knowing is part of executing well.

2. The Banker: funding, resourcing, and buying time back

Strategy without resources is cosplay. KM needs a budget that matches the ambition: infrastructure, tooling, community enablement, facilitation, and the unglamorous work of curation and stewardship. Approving a reasonable budget is not generosity; it is the price of admission.

Time matters even more than money. KM leaders can’t lead KM “after the real work.” If leaders aren’t empowered to share and learn, the program suffocates under meetings, delivery pressure, and a thousand urgent priorities. The same is true for everyone else: if the organization doesn’t protect time for sharing, innovating, reusing, collaborating, and learning, KM becomes performative. Leaders build KM capacity by protecting slack, real calendar space, so reflection and reuse can occur without penalty.

3. The Advocate and Protector: banishing fear and wiring in accountability

KM requires policy and practice, space and technology, the levers of culture, to accept it, permit it and facilitate it. People won’t share what they suspect will be weaponized, and they won’t admit lessons learned if lessons are treated as confessionals. Senior leaders set the tone: psychological safety can’t just be a slogan, or even a policy; it must be a living practice that gets rewarded.

Protection has two concrete faces: accountability and recognition. KM behaviors can’t be optional virtues. Leaders embed KM goals into performance expectations for all employees, then inspect them the way other goals are inspected. When leaders track revenue goals rigorously but treat knowledge goals as aspirational, the organization learns what matters.

Reward closes the loop. Leaders advocate for the people who share, reuse, mentor, and improve the commons. Public and private recognition teach faster than policy. Incentives don’t need to be extravagant, but they do need to be consistent and visible.

4. The User: leading by example on the platform

KM platforms become meaningful when executives behave like users, not landlords. If the top team never logs in, comments, asks questions, or posts, the platform becomes “for us,” not “for them” (leadership), which punctures the balloon of motivation.

When leaders use the KM system to run work, like asking questions in a forum, pointing to reusable assets, commenting on drafts, and acknowledging contributions, the participation of others follows.

This is not “being social,” it’s being observable. Leaders who show up inside the knowledge environment normalize the behavior they want: curiosity, transparency, and reuse over reinvention. They recognize learning as an asset and not a cost.

Don’t just view KM from above

Leadership only earns the title of architect, banker, and user of knowledge when it stops admiring KM from the balcony and starts working in the commons: showing up in the system, spending real time and budget, asking questions in public, and giving direct reports time, respect and acknolwedgement for their knowledge-realated activities, which ultimately results in the organization treating stories, lessons, and reusable assets as core infrastructure.

KM Leadership in a world with AI

I didn’t integrate KM into this analysis. All of the points made in the main post apply regardless of whether AI is used or not. The role of leadership doesn’t change, regardless of the tool. Should leaders encourage the safe and ethical use of AI? Of course they should. But good leaders also espouse safe working conditions in general. If they incorporate these ideas into their management practices, they can anticipate the change AI is bringing about. A good knowledge management program is not just about lessons learned and the sharing of good practice, but also about the management of learning, which is the key to accepting, naviagating and integrating change.

AI will likely transform many aspects of how an organization implements knowledge management. What won’t change is why organizations should implement KM, or the non-technological components of program success. If AI evolves in the way people think it will, toward a capability that answers questions directly, and reconfigures systems based on data, it will give people time to improve on the people, process and social capital aspects of knowledge management: areas where productivity gains can be measured in days and weeks, not just minutes.

AI-empowered Knowledge Management can be a strategic tool for achieving goals and objectives, but only for organizations that are clear about their goals and objectives, and that count knowledge among their strategic assets.

1. Anchor KM (and AI) in Business Strategy

KM only matters if it amplifies strategic intent. AI only matters if it does the same.

I start by asking leadership to describe the organization’s strategy in concrete, behavioral terms. Not “be innovative,” but “cut time-to-launch in half” or “raise renewal rates in key segments.” I want to know what they are trying to achieve and which obstacles they plan to overcome. KM and AI should sit underneath those ambitions as enabling systems, not as independent agendas.

Different strategic postures imply different knowledge and AI priorities:

• Operational Excellence

  KM and AI focus on reducing friction: clean processes, codified best practices, and cross-training that lowers delivered cost. AI can automate routine decisions, surface standard operating procedures in context, and monitor process variance.

• Product Leadership

  The objectives lean toward accelerating discovery and commercialization. KM creates reusable research, design rationales, and experiment logs. AI supports this with generative tools for concept exploration, pattern finding in R&D data, and simulation that reduces the cost of failure.

• Customer Intimacy

  The center of gravity is a shared understanding of customers. KM curates narratives, journeys, and account histories. AI augments this with synthesized customer insights, segment-level pattern recognition, and copilots that help teams personalize responses without starting from scratch each time.

KM and AI strategy should be written as a derivative of business strategy, not as a parallel universe.

2. Surface Business Pain, Not Just KM Aspirations

I suggest that the next level of analysis starts with a capability map where the organization exposes not its organizational structure, but its operating assumptions. After strategy, exploring how work gets done reveals the pain points in search of relief. KM and AI should go to work where time, money, and trust are already leaking.

I look for signals such as repeated reinvention of deliverables, slow onboarding, stalled decisions, interrupted product launches, specific employee dissatisfaction with how work gets done, or communication that stalls because “the person who knows is busy,” or customer issues that echo across regions. Those become areas of hypothesis for KM and AI.

Talking with key roles sharpens the list:

• CEO – Does the organization know what knowledge differentiates it, and can that knowledge be mobilized at speed? AI highlights how often employees re-ask the same questions or rebuild the same artifacts.

• CIO / CDO – How fragmented are content, data, and systems? Where is AI already appearing as shadow IT because people can’t get what they need from official tools?

• HR / Learning – How are skills and expertise mapped, and how quickly can teams find people who have done something similar before? AI can help infer skill graphs from project histories and content, but only if those are being captured.

• Marketing / Sales / Service – How is customer knowledge captured, reused, and challenged? AI shows up here through copilots, summarization, and recommendation—but it will only be as good as the underlying KM practices and data.

The objective is to map concrete pains to the kinds of knowledge and decision flows that would relieve them.

3. Translate Business Needs into KM & AI Services

Business language rarely arrives preformatted as KM requirements.

A statement like “we need answers faster in the field” needs to be unpacked into specific services: expertise location, validated FAQs, decision trees, playbooks, maybe an AI assistant that fronts them, along with a knowledge graph that appreciates the semantics of customer interactions.

Most needs fall into a familiar set of KM drivers:

• Capturing critical knowledge

• Collaboration and communities

• Knowledge transfer and learning

• Expertise location

• Knowledge analysis and insight

• Search and findability

AI can sit inside each driver:

• Expertise location enhanced by inferred skill graphs and smart routing.

• Search backed by retrieval-augmented generation (RAG) over curated content.

• Knowledge analysis using machine learning to detect patterns in projects, failures, or customer feedback.

The work here is to map pains and ambitions into a small number of KM and AI services that can be named, designed, and owned.

4. Run a Gap and Opportunity Analysis

Once needs are clearer, the next move is to compare “what we have” with “what we need.”

• Knowledge Mapping

  Map where critical knowledge lives: in people, documents, data, workflows, and external partners. This can be done as a classic map on a wall, but AI now enables mining repositories, chat logs, and ticketing systems to infer assets, topics, and connections. That doesn’t replace human validation, but it accelerates discovery.

• Opportunity Analysis

  Identify places where the organization is already strong—proprietary methods, unique datasets, distinctive practices that could be scaled. AI often magnifies these strengths through reuse: codifying a method into a playbook + copilot, or exposing a distinct dataset to analytic and generative tools.

• Readiness Review

  Cultural readiness still matters more than technology. If people don’t share, don’t trust, or don’t have time, neither KM nor AI will fix that on their own. A realistic assessment of leadership sponsorship, incentives, and change capacity helps avoid ambitious designs that will die on contact with the lived work experience.

Gap analysis is where ambition meets constraint in a useful way.

5. Prioritize by Value and Feasibility

No organization can “manage all the knowledge” or “apply AI everywhere.” Nor should it try.

The focus should be on strategic knowledge: the flows and assets that actually change outcomes when better captured, shared, or analyzed.

To prioritize:

• Filter Options

  Screen KM and AI ideas against criteria: alignment with strategic goals, value if successful, resource feasibility, technical feasibility, and likelihood of acceptance. Some ideas are right, but not now.

• Impact vs. Difficulty

  Use a simple matrix: high-impact / low-difficulty items go first. AI often shines in these early wins with automated classification, better search, and content summarization, because they lower friction without demanding a full redesign of work.

• High-Leverage Domains

  Focus early KM and AI investments where disruption is expensive and learning compounds: competitive intelligence, customer care, safety, compliance, innovation portfolios.

• Understand Constraints

  If the organization is capacity-constrained (more demand than it can handle), KM and AI should reduce internal friction, improve reuse, and lift productivity. If it is market constrained (more capacity than demand), objectives should lean toward better market sensing, sharper targeting, and faster, more tailored offers.

Prioritization is where the wish list becomes a program.

6. Define Critical Success Factors and Metrics

Critical KM objectives need to be translated into a small set of measurable Critical Success Factors (CSFs). AI doesn’t change the need for measurement; it changes what can be measured and how quickly.

Examples:

• For customer intimacy:

  CSFs might include depth and coverage of customer profiles, the percentage of engagements informed by prior insights, or the reduction in time to respond to complex requests. AI can help log and analyze interactions to infer whether prior knowledge was actually used.

• For operational excellence:

  CSFs might include reuse rates of standard assets, cycle times, error rates, and variance across teams. AI contributes by detecting process anomalies and suggesting relevant assets in the flow of work.

• For product leadership:

  CSFs might include time from idea to experiment, experiments per quarter, and the proportion of decisions informed by prior learning. AI can speed literature review, design exploration, and pattern detection, but the metrics should still be grounded in how fast and how well the organization learns.

Measurement frameworks such as a Balanced Scorecard remain useful, now augmented with telemetry from AI systems and KM platforms. Leading indicators (engagement with knowledge assets, AI-assisted tasks completed, reuse rates) should sit alongside lagging indicators (margin, growth, satisfaction, risk reduction).

Suggested Next Step

Once critical KM and AI objectives are clear, the practical next move is to design a simple measurement and learning framework around them. Pick a handful of CSFs, define leading and lagging indicators, and connect them to specific KM and AI services.

That creates a feedback loop: strategy → KM/AI services → behavior and performance → evidence back to leadership. Without that loop, KM and AI both risk becoming experiments in search of a purpose rather than instruments of strategy.

The success of KM initiatives, whether focused on reuse, collaboration, or innovation, hinges fundamentally on people. While technology provides the essential scaffolding to manage content and make it accessible, human roles define the practices and capabilities required not only to create new knowledge, but to capture and transfer that knowledge in context.

Organizations must resist the urge to deploy technology first and then search for people to run it later. Instead, a careful definition of roles and responsibilities must align with the strategic intent of AI-powered KM, embedding knowledge work into the flow of day-to-day business. We must look beyond merely creating new titles and focus on the tasks that elevate the organization’s competence and capacity for learning.

Here are the primary roles an organization should consider when structuring a robust knowledge management program, acknowledging that many of these tasks can and should be integrated into existing job descriptions:

Chief Knowledge Officer (CKO)

The existence of a CKO is not necessary for every organization, particularly those where knowledge sharing is already an organic and embedded process. However, for organizations whose product is knowledge, or those needing a strong catalyst for knowledge-oriented change management and a reimaging of knowledge-related processes and practices, the CKO is essential.

The CKO is not the owner of organizational knowledge; rather, they serve as the catalyst for empowering individuals, teams and functions with knowledge-sharing practices and policies. Their responsibilities are highly strategic and operational, including:

• Knowledge Environment Design: The CKO should be the chief architect of the knowledge environment, owning the specifications for infrastructure, content design (e.g., best practices and lessons learned), and the tools for converting tacit knowledge into explicit knowledge. In today’s AI-driven world, the CKO and the CAIO, or other AI leader, should synchronize their efforts to unleash organizational knowledge.

• KM Integration: They collaborate with other process owners to integrate KM tasks, such as sunset reviews on projects and the use of collaboration spaces, into core business processes.

• Value Reporting and Gap Elimination: They function as KM’s analyst and reporter, creating knowledge metrics tied to strategic goals. They identify knowledge gaps that may necessitate training, hiring, or even the acquisition of other firms.

• Organizational Transformation: They must be a leader who can overcome internal politics, move the organization toward knowledge sharing, and promote the adaptive, flexible, and evolutionary models required for AI-first businesses.

The CKO should be a senior manager and a member of the senior management team, not merely a department head in IT or library services.

Knowledge Stewards and Content Managers

This role concentrates on the local ownership, life cycle, and quality control of codified assets. We highly encourage the development of Knowledge Stewards, often closely associated with process ownership. With AI’s reliance on unstructured content, Knowledge Stewards can and should act as the front-line leaders in preparing legacy content for AI use.

• Knowledge Steward: Encourages the capture and revision of knowledge. In a Communities of Practice (CoP) context, they actively nurture the CoPs, helping members transform their knowledge into usable forms and bridge boundaries between communities.

• Content Manager: Responsible for organizing and distributing information based on the interests and needs of the user community. This includes cataloging content, determining what critical knowledge should be contributed to repositories, and ensuring a rigorous review process to make only the highest-quality content available. They often act as an “informed layer” between information specialists and content experts.

Knowledge Brokers and Facilitators

These roles are crucial for maximizing knowledge transfer, particularly tacit knowledge, which cannot be easily codified.

• Knowledge Broker: Connects knowledge seekers directly to sources of tacit expertise. This function is vital in organizations, often performed informally by individuals who know who knows what, such as Librarians.

• Facilitator: Maintains discussions, keeps dialogue business-appropriate, and helps manage the visible aspects of community events. In a CoP structure, facilitators are often part-time responsibilities of key employees and are critical for ensuring user questions are answered.

Taxonomists and Information Architects

These specialists provide the structure necessary for efficient content discovery and access.

• Taxonomist: Creates enterprise taxonomies applied by content managers or automated tagging systems. They are indispensable for moving beyond simple portals to creating more comprehensive representations of knowledge and relationships within the corporate data model.
  
  In an AI world, taxonomists are responsible for creating and maintaining the ontologies and other representations necessary to power knowledge graphs and graphRAG implementations.

• Information Architect: Focuses on packaging knowledge to ensure that content is consistent, meaningful, and labeled in a way that is relevant to the end-user, often arbitrating across knowledge functions (e.g., ensuring marketing information is usable by sales personnel).

Knowledge Engineers

This role traditionally focused on complex systems that involve artificial intelligence (AI). A knowledge engineer is skilled in writing rules, developing cases for Case-Based Reasoning (CBR) systems, and coding reasoning algorithms. They are also skilled in knowledge acquisition techniques necessary for extracting deep semantic knowledge, a difficult task.

Some might argue that knowledge engineers have become obsolete in the era of generative AI, much of the governance of Large Language Models requires rules and predicate logic to, for instance, frame guardrails. Knowledge Engineers may no longer be at the forefront of mapping expertise, but they are crucial to ensuring the safety of modern AI implementations.

Integration: Making Knowledge Management the Way We Work

Focusing on these key roles should not lead to the creation of new organizational silos. Knowledge management is intrinsically intertwined with the core functions of the business: People, Process, and Technology.

The ultimate goal is for knowledge capture and sharing to become seamlessly embedded in daily business processes, making it a natural part of the workflow, rather than a separate task or “add-on” activity. That aligns with a fourth core function: encouraging connection and social context.

KM roles must integrate well with Human Resources (HR) for competency mapping, performance evaluation, and the development of learning strategies. They must partner with IT to ensure the technology architecture supports common tools and processes for sharing and collaborating, rather than proliferating competing, fragmented solutions.

Finally, strategic integration requires that CKO functions, such as knowledge gap elimination and value reporting, are closely aligned with the strategic planning and operational implementation of business unit leaders.

By defining KM roles based on function and then embedding these functions across the organization, we move toward a state where every employee is recognized as a knowledge worker/knowledge manager, which is the necessary prerequisite for achieving competence, adaptation, and sustained innovation.

Digital transformation promises much but often delivers less than expected. It seems almost silly in the era of AI to discuss digital transformation anymore, as most organizations have likely already undergone digital transformation. They should now focus on retransformation.

Of course, we have been through a “re” before, during the reengineering revolution. I wrote then that it was impossible to reengineer something that wasn’t engineered in the first place. We have moved beyond that. Organizations have become overwhelmingly digital, but most have not become effectively digital.

Organizations treat technology as if it alone will realign work, shift company policies and practices, and create value. That organizations are still talking about digital transformation is evidence that it has become just a lifecycle metaphor for tool acquisition layered on top of old habits. The rhetoric still suggests inevitability, but effective use of technology requires discernment and judgment.

Knowledge Management (KM) should sit at the center of this tension. KM has always insisted that people, processes, and practices anchor transformations, and that perspective remains critical. Technology, the new and shiny, the easily acquired, often outweighs those considerations in transformation programs. KM’s practices for reflection and collaboration can help organizations discover how to become effective at using technology, even if they aim to accomplish great things at great speed.

KM becomes a steadying force, perhaps seen by some as a throttle, but I see it more as a moment to seek reason when speed leans toward the edge of chaos, and uncertainty becomes context.

The need for rational reflection has not shifted simply because AI and machine learning are now part of the mix. If anything, the arrival of generative AI reinforces KM’s original point: without governance, context, and trust, AI only magnifies noise. The promise of transformation without policy and practice alignment or thoughtful integration is the reason even those who have digitally transformed continue to discuss digital transformation rather than strategic alignment, execution and navigation.

Effectively digital means leveraging technology as a tool to execute strategy, not as a topic that requires a strategy.

Rethinking Findability in the Age of AI

AI introduces new dynamics to our relationship with information. Intelligent search, summarization, and recommendation systems can accelerate access to knowledge, yet they are built on fragile foundations. Without robust taxonomies, metadata, and curation, AI models are left to infer meaning from unstructured sources. They may generate plausible answers that dissolve under scrutiny.

Knowledge graphs illustrate both potential and risk. When well-designed, they create a semantic web that links structured and unstructured information. When poorly managed, they become brittle, encoding outdated assumptions that AI will repeat without question. KM must treat these tools not as replacements for human sense-making, but as amplifiers that require active stewardship.

Collaboration Beyond the Platform

Remote and hybrid work have made collaboration platforms essential; however, the temptation is to view them as ends in themselves. KM’s role is to ensure these environments support authentic communities of practice, enable expertise location, and capture tacit insights.

AI can enhance collaboration by surfacing expertise and recommending connections, yet it can also distort organizational dynamics if treated as an oracle of truth. A recommendation engine may privilege the already visible and marginalize the less documented but equally valuable knowledge.

At the same time, AI introduces new collaborators, and I say that as a plural, because the effective use of AI may not be a single system for analysis or insight, but leveraging multiple systems as adversaries to challenge and explore the edges of ideas that may fall out by humans working in stressful work environments.

The deeper question is how much organizations want to delegate social connection to algorithms. The answer should be pretty clear. While AI can mirror the codified, it can’t gain access to what people haven’t shared, or are just in the process of learning, without people to facilitate the transformation of data, experience and context into knowledge. KM recognizes that meaningful collaboration depends on trust, relationships, and human choice; and only meaningful collaboration leads to collective knowledge.

Embedding KM in Process

The best test of KM lies in process integration. Knowledge practices must become embedded in daily workflows, rather than being appended as afterthoughts. AI assistants that automate routine knowledge tasks—such as drafting reports, tagging documents, and summarizing conversations—can add significant value. But they also introduce opacity. If AI-generated insights influence decisions, organizations must demand transparency, provenance, and the ability to interrogate the logic behind recommendations.

This skepticism should not be read as dismissal. The efficiency gains are real, but without accountability, automation risks eroding trust rather than building it. KM must frame AI not just as a convenience, but as part of a governed knowledge ecosystem.

The Limits of Transformation

Digital transformation is often presented as a destination. In practice, it is a continuous negotiation between technological affordances and organizational realities. Change fatigue is real, and workers respond best when they see direct, personal benefit. KM practitioners are experienced in articulating this benefit because they have long wrestled with the “what’s in it for me?” question.

The language of transformation implies finality, turning one thing into something else, but because technology continues to evolve, transformations likewise need to remain continuous.

There is no future proofing, and technological maturity is a myth. Maturity models and end-states suggest stability that rarely exists in practice. A more honest approach is to recognize digital change as a series of experiments, adjustments, and accommodations —a pattern of adaptation rather than a linear march.

And keep in mind that the arrival of new technology doesn’t cause the need for digital transformation; instead, it creates new expectations, informed and driven by social, economic, environmental, and political factors.

Grounding the Future

Knowledge remains a durable asset. Technology, whether enterprise resource planning, resume scanners, collaboration platforms, or AI agents, should be judged by how well it preserves, shares, and extends that knowledge.

The task is not to replace human intelligence with machines but to design systems that support collective sense-making. AI is best seen as a partner in that effort, one that demands human oversight to remain useful.

Digital transformation proves less about radical reinvention than about creating conditions for knowledge, individuals, social networks, and practices to thrive amid shifting circumstances.

Organizations that succeed will be those that accept the limits of tools, invest in policy and practice, and treat knowledge as both a process and an asset, and recognize that they aren’t seeking to be transformed, but rather that they are living in a constant state of transformation.

Blending agile methods with AI in KM requires policy and practice shifts, alongside innovative technology adoption. Here are some practical good practices to guide your journey, informed by real-world successes:

1. Make Knowledge Everyone’s Job (Cross-Functional Teams): Don’t leave KM to a small, siloed team. Create a cross-functional knowledge structure or working group, or even better, communities of practice accountable for knowledge that embrace diverse disciplines, such as IT, support, operations, and HR, who generate or use critical knowledge. This mirrors the agile principle of cross-functional teams and ensures diverse input.
   
   Only smaller organizations should consider a single knowledge management team. Larger firms need to consider knowledge management in terms of function and geography, allowing communities to form that support knowledge in a distributed manner, with like functions across geographies coordinating their efforts.

2. Treat Knowledge as a Product, with a Backlog and Sprints: Just as software teams maintain a backlog of features and fix bugs in iterations, maintain a knowledge backlog. Capture requests for new articles, needed updates, and ideas for improvements in a tool or spreadsheet that the team continually prioritizes and updates. Use short iterations (sprints) to tackle a set of these items, then release the updated knowledge (publish those articles, refresh that FAQ) immediately.
   
   A backlog also helps manage the contribution of AI: for example, if the AI flags ten outdated documents, they can be added to the queue and allocated, a few at a time, during each sprint. A telecom company broke their policy manual overhaul into bi-weekly sprints, each cycle updating and releasing a handful of policies. Within a few months, their entire repository was refreshed without a massive project upfront, and it stays up-to-date through continuous small increments.

3. Leverage AI as Your Knowledge-Assistant: Incorporate AI tools to assist, not replace, your knowledge workers. For example, enable support agents or technical writers to use an internal generative AI assistant when creating or updating articles. This might be a bot integrated in the knowledge management system that can draft an article based on a support ticket or suggest relevant content to include. (See number 8 below for precautions and guidance.)
   
   By doing this, organizations reduce the effort required for documentation and dramatically speed up content creation. However, always have the human expert review the AI’s suggestions; this not only ensures accuracy, but also turns the process into a learning opportunity (the human might catch errors and feed corrections back into the system, training it over time, and the person might see patterns across their reading that the AI misses, such as a need to restructure a problem solving approach that has become less effective over time).
   
   Think of it like pair programming, but for documentation: the AI writes a first pass, the human refines and approves, and both learn.

4. Integrate Knowledge Capture into Workflows: Agile KM works best when creating or updating knowledge isn’t a separate, cumbersome task outside daily work. Follow the principle of capturing knowledge in the moment. That means that service desk software, project management tools, or collaboration platforms should make it easy to contribute to the knowledge base without disrupting the workflow.
   
   For instance, if a consultant finds a new insight during a project, they should be able to jot it down in minutes, with a click, from whatever app they are using. Some teams implement lightweight forms or chatbots (“Did you just solve a problem? Briefly note the solution here…”) to prompt people to contribute on the fly.
   
   Easy interface = more knowledge captured. Modern AI-enhanced systems even allow contributions via voice or chat – an engineer could tell a bot the solution they just found (or dictate it while driving), and let the AI transcribe and draft it into a shareable format. The key is removing friction from knowledge capture.

   

5. Use AI to Automate Housekeeping: Over time, knowledge bases can become cluttered with duplicate, irrelevant, or outdated entries (the dreaded knowledge ROT: redundant, obsolete, trivial content). AI can be the cleanup crew. Set up AI tools to periodically scan the repository for anomalies, such as multiple articles addressing the same question (suggesting they be merged, even how), content that hasn’t been viewed in over a specified period of time (which may imply it is obsolete), or inconsistent terminology that could confuse users.
   
   Some organizations schedule a monthly “AI audit” report that flags potential issues identified by AI. The knowledge team or community reviews that content and decides what to do (keep, update, or delete). This ensures continuous quality without relying solely on someone remembering to review old pages. These AI-driven audits can save hundreds of hours by pinpointing which knowledge articles need attention.

6. Measure and Adapt: In agile, we inspect and adapt. Similarly, establish metrics for KM efforts and let the data guide improvements. Track usage stats: which articles are most viewed, which searches yield no results, and what questions are users asking the chatbot that it can’t answer. Also track outcome metrics, such as support case deflection rates, average time to find information, and employee onboarding time reduction, among others.
   
   For example, after implementing agile KM practices plus an AI search assistant, a company might see that employees are finding information 40% faster (measured by time-to-answer in the service desk) or that new hires reach full productivity in 2 weeks instead of 4.
   
   If a metric isn’t moving in the right direction, use agile problem-solving, do a root cause analysis, experiment with a change in the next sprint, and see if it improves.
   
   Metrics not only record the value of KM (to secure continued funding and support), but they also highlight where the approach needs improvement. [Note: This aligns with my position that all “process” work is knowledge work. The process captures “process” knowledge. Therefore, iterating on a process is no different than iterating on content, save for the need for more robust change management practices.]

7. Cultivate Knowledge-Sharing (with Leadership Support): No method or tool will succeed if the organization doesn’t value knowledge. Leaders should encourage teams to dedicate time to documentation, celebrate those who contribute valuable insights, incorporate knowledge goals into performance reviews, and regularly share their knowledge and how they apply organizational knowledge. In my experience, organizations that excel at KM treat knowledge as core infrastructure, as important as their products and services. They move beyond the mindset of “knowledge is extra work” to “knowledge is part of everyone’s job.”
   
   A practical tip: hold regular knowledge-sharing sessions (such as brown-bag sessions or “demo days” where teams showcase new knowledge they’ve captured or a cool use of the AI tool to solve a problem). This normalizes knowledge work and generates enthusiasm.
   
   Also, ensure incentives are aligned – if support agents are only rewarded for closing tickets quickly, they won’t take the extra 5 minutes to document solutions. Adjust KPIs to reward knowledge contributions.

8. Governance and Guardrails for AI: Establish guidelines so AI remains a helpful partner, not a loose cannon, by setting up a review workflow for AI-generated content.
   
   Maintain transparency about AI’s sources: if an AI-generated answer draws from a specific document or external site, have it cite that source so users can verify the information. Even better, have it provide the answer in context. This builds trust in the content. Be mindful of information security. Those who employ cloud AI services must ensure they’re not feeding sensitive data without proper encryption or agreements. And importantly, educate staff about AI, its capabilities and its limitations.
   
   When everyone understands that the AI might occasionally err or “hallucinate,” they’ll be more vigilant, and the whole organization will use the tool more effectively. A balanced governance approach will enable organizations to harness AI’s speed safely, thereby maintaining the credibility of the knowledge base.

By following these practices, organizations can create a synergistic system: agile methods ensure continuous improvement and relevance, while AI tools provide scale, speed, and intelligence. Companies that get this right are effectively building a self-learning organization, one that quickly absorbs new information, disseminates it through intuitive AI-powered channels, and adapts its knowledge stores in response to real-world usage.

Knowledge management is in flux. The forces of artificial intelligence, hybrid work, and shifting expectations between employers and employees are reshaping how organizations think about knowledge. Add to that the constant fight against misinformation, the glut of collaboration tools, and the rise of mobile and AI brought in by employees themselves, and the KM discipline finds itself navigating a profoundly altered landscape.

AI-Powered and Generative Knowledge Management

Artificial intelligence should no longer be considered experimental in KM; it has become foundational. Generative AI, semantic search, and large language models now handle the chores of classifying and tagging documents or highlighting outdated material. Machine learning automates repetitive curation, allowing humans to focus on sense-making.

Natural language processing enables contextual search, not just keyword retrieval. Generative systems summarize reports and create draft knowledge articles. Integrated chatbots trained on internal repositories deliver contextual answers 24/7, offering employees and customers a faster path to insight.

This isn’t just efficiency. AI-powered analytics now mine organizational knowledge for patterns, trends, and anomalies that would otherwise remain undetected manually. The result is knowledge that is more accessible, more dynamic, and increasingly central to organizational agility.

Hybrid and Distributed Work Knowledge-Sharing

Hybrid work has turned knowledge into a distributed asset. With over a third of employees working remotely full-time and nearly half in hybrid models, the need for centralized, always-available knowledge has never been greater.

Conversations that once happened in hallways now depend on systems. Intranets, wikis, and integrated AI assistants embedded in platforms like Slack or Microsoft Teams bring knowledge into the workflow. A person in London can post a decision at the end of their day, and a colleague in Seattle can act on it the next morning.

Asynchronous sharing is becoming standard. Meeting recordings, discussion threads, and decision logs help people stay aligned despite time zones. Hybrid work makes cloud-based, user-friendly KM essential—it is the connective tissue holding distributed teams together.

Changing Workforce Dynamics and Knowledge Retention

Many workers don’t stay long. The median number of years that wage and salary workers had been with their current employer was 3.9 years in January 2024 (BLS, Employee Tenure 2024). The “job for life” model is gone, replaced by frequent turnover and contingent labor. That means knowledge retention is no longer optional—it’s a survival tactic.

Organizations are formalizing capture at both entry and exit. Onboarding now includes pulling tribal knowledge from veterans, while offboarding is as much about capturing insights as it is about returning badges and laptops. Mentoring, documentation, and recorded processes have become critical safeguards.

Freelancers and gig workers also challenge KM boundaries. Their contributions must be captured and shared, but access and permissions must be managed carefully. Digital-native generations, meanwhile, expect intuitive, consumer-grade tools. If internal systems don’t meet expectations, they’ll turn to shadow IT—using unauthorized apps and devices.

Evidence-Driven Work and Data-Informed Knowledge

Evidence-based decision-making has become the new standard. Analytics and knowledge systems are converging, with AI-driven tools trawling unstructured documents and correspondence to surface patterns that inform business decisions.

Not all data matters equally. Smart curation, supported by AI, focuses on quality and relevance, pruning outdated or low-value information. The result is higher trust in organizational knowledge.

Integrations with BI dashboards and research libraries ensure that decisions are backed by both internal knowledge and external data. APQC emphasizes that KM is now inseparable from digital transformation and evidence-based practices.

Combating Misinformation and Ensuring Trusted Knowledge

The battle against misinformation is not confined to politics or media—it seeps into the workplace. Employees bring in unverified data, or worse, AI-generated hallucinations dressed up as facts. KM systems must protect against this erosion of trust.

Verification, expert review, and single sources of truth are rising priorities. Generative AI can accelerate access to knowledge, but without human curation and validation, it risks polluting trusted repositories. Training employees in information literacy is now an essential part of knowledge management practice.

New approaches are also emerging that go beyond traditional review. Knowledge provenance and traceability are becoming core requirements for AI-enabled KM. Retrieval-Augmented Generation (RAG) systems are being paired with strong document management practices, ensuring that AI answers are grounded in verifiable sources rather than opaque inference. Tools that provide explicit citations, source links, and contextual metadata help employees trust what they see.

Provenance frameworks extend this further. Metadata knowledge graphs record data lineage—what was used, when, by whom, and how it was transformed—making each knowledge artifact auditable. Research efforts, such as full traceability models for knowledge graphs, even track changes at the triple level with timestamps and authorship records. Emerging agentic AI standards, like PROV-AGENT (see PROV-AGENT: Unified Provenance for Tracking AI Agent Interactions in Agentic Workflows), capture not just outputs but also prompts, workflows, and decision paths, offering transparency into how multi-agent AI systems produce recommendations.

The World Economic Forum identified disinformation as one of the top global risks in 2024 (WEF, 2024). Organizations are applying that urgency internally, making knowledge validation, provenance, and traceability core functions of KM. The goal is simple: every answer should come with a pedigree that employees can interrogate, verify, and trust.

Navigating Collaboration Tool Overload

Collaboration sprawl is real. Enterprises often juggle dozens of platforms—Slack, Teams, Notion, Miro, SharePoint, email—each capturing slices of knowledge. The result is silos, duplication, and wasted time as employees hunt across systems.

This fragmentation has become a strategic concern. Organizations are moving to consolidate, simplify, or at least unify search across disparate platforms. Enterprise search engines that index multiple repositories are in high demand.

Vendors like Microsoft and Slack are positioning their tools as integration hubs, but open standards and federated approaches are also regaining attention. Without rationalization, the very tools designed to foster collaboration risk undermining it.

See more on collaboration tool overload in the Serious Insights report, Why Collaboration is Broken: Becoming Anti-Fragile Through Design.

Mobile, BYOD, and Bring Your Own AI

Knowledge has gone mobile. With most searches now happening on mobile devices, KM platforms are expected to be accessible anywhere. Frontline and field workers rely on phones and tablets to find instructions, guides, and manuals in the moment.

But the bigger frontier is AI. Employees aren’t waiting for IT approval—they’re bringing ChatGPT, Claude, and other generative tools into their work, often unsanctioned. This “Bring Your Own AI” mirrors the BYOD movement a decade ago.

The risks are substantial: data leakage, compliance violations, and exposure of proprietary information (See, Bring Your Own AI: How to Balance Risks and Innovation, MIT Sloan Management Review, 2024). Banning will prove futile except in the most secure of environments. Many organizations expect that most employees will use AI tools by 2024, regardless of policy. The push to remain competitive and employed creates new information risks.

Forward-looking companies are embracing governance frameworks and providing enterprise-grade AI assistants, often embedded in platforms like Microsoft 365 Copilot and Salesforce. Others, like Unily, are enabling secure multi-agent environments that manage different AI tools under enterprise policies.

The Near-Future of KM

Knowledge management is being reshaped into a more adaptive, intelligent, and employee-centered discipline. AI is accelerating curation and access. Hybrid work is pushing knowledge to the cloud and into the flow of daily tools. Workforce churn and gig labor are forcing better ways to capture and retain knowledge.

Misinformation demands vigilance, while collaboration overload requires simplification. Mobile access and Bring Your Own AI make KM more personal, but also more complex. The organizations that succeed will be those that proactively embrace these shifts—building trust, securing knowledge, and integrating intelligence into how work gets done. This involves balancing tools with policy and practice, and finding the right balance that works for them.

AI is also moving the tech stack beyond a one-size-fits-all world into a world of adaptive and responsive applications that will likely evolve rather than being designed. Knowledge of the business, its markets and its customers will shape those future operating models and the applications that support them. Organizations will need to focus as much on awareness of emergent needs as on codifying past lessons and historical knowledge.

Visit Serious Insights

Knowledge management (KM) in many organizations has long suffered from a waterfall mentality – capturing knowledge in big, slow batches and storing it in static repositories. The result? Outdated FAQs, stale intranet pages, and frustrated employees who can’t find the insights they need. Today’s fast-paced environment demands agility, and a tight partnership with AI and search that can illuminate what is essential, focusing attention on the content that needs curation.

As an industry analyst, I’ve seen firsthand how agile methods combined with artificial intelligence (AI) are breathing new life into corporate knowledge practices. In this post, I’ll break down how these two ideas intersect and provide practical steps (with real examples) to make an organization’s knowledge management more dynamic and effective.

From Waterfall to Agile Knowledge Management

Traditional KM often followed a linear, “publish and forget” model, much like waterfall software projects. Content would be created, polished, approved, and finally published, often weeks or months after it was created, with little understanding of how it might be used again in the future, if at all.

As the requirements for knowing change, documented knowledge often becomes obsolete. In some cases, it may retain value, but in a different context from its original creation intent.

Agile knowledge management seeks to introduce flexible and iterative processes that continuously adapt, emphasizing quick updates, collaboration, and learning through hands-on experience. Instead of treating knowledge as a one-time deliverable, agile KM treats it as a living product that is constantly refined.

Real-world innovators have embraced this shift. For example, Google and Amazon encourage agile knowledge-sharing practices, fostering a culture where employees freely share insights and iterate on ideas – a practice credited with helping to seed many groundbreaking products and services.

In the financial industry, firms have learned that providing decision-makers with real-time access to up-to-date market knowledge (rather than waiting for monthly reports) yields faster and more informed investment decisions. The common thread is agility: breaking down silos and accelerating the flow of knowledge (and data) across the organization.

Adopting an agile approach requires designing processes and roles. In Scrum terms, think of knowledge as the product: it has a backlog, a dedicated team (a community of practice), and frequent incremental releases. Teams across industries, from tech to aerospace, report that agile KM not only keeps information current but also sparks innovation by enabling cross-functional collaboration.

Making KM an agile practice (instead of a rigid process) aligns perfectly with modern, resilient operations that seek to eliminate waste and accelerate innovation. In other words, agility in KM isn’t just a nice-to-have; it has become a must-have for organizations that want to leverage knowledge as a competitive advantage.

Case in Point: Agile Knowledge Management in Action

How does agile KM look in practice? Let’s consider a real example from a contact center setting. Implementing a new knowledge base for customer support, with hundreds of articles, can feel overwhelming. One service organization tackled this by applying Scrum-style project management to their KM initiative :

• Cross-functional Team: They formed an agile knowledge team comprising support agents, subject matter experts, technical writers, and managers, ensuring that all perspectives (from frontline users to content approvers) were represented. This diverse team structure ensured that the knowledge base was built with a 360-degree view of what information was truly needed and how it would be utilized.

• Clear Objectives & Metrics: At the outset, the team identified pain points (e.g., high call escalations on specific topics, lengthy training times for new agents) and established measurable goals for the knowledge project. For instance, they aimed to reduce “I can’t find info” support tickets by X% and improve the first-contact resolution rate. Baseline metrics were captured to enable the quantification of improvements later.

• Backlog of Knowledge Content: Instead of a giant documentation marathon, they created a prioritized backlog of knowledge articles to write or update. High-impact items, such as addressing the most frequent customer issues and known product glitches, were tackled first. Less urgent topics are queued behind. This backlog was groomed continuously, just like a software backlog, to reflect changing business needs (for example, when a new product launched or regulations changed, related knowledge items moved to the top of the list).

• Iterative Sprints: Work was divided into short sprints (in this case, two-week cycles) with specific deliverables, for instance, 20 new or updated articles per sprint. At the end of each sprint, the team delivered those knowledge assets into the repository for immediate use.
  
  This sprint cadence brought several benefits: quick wins, continuous momentum, and early feedback on whether agents found a helpful article (or not), which was looped in before the next sprint.
  
  I have heard from many agents that the amount of information is just overwhelming. Assigning ownership and priorities, when everything is on the table, becomes “too big a job.” That second-tier immensity often leads to inaction rather than accountability. Agile sprints help create actionable, consumable moments. While the totality of the content repository may remain overwhelming, this approach focuses on what’s most important, allowing demand to create a window into a narrow passage of the repository, rather than forcing everyone to see the entire warehouse.

• Daily Stand-ups & Collaboration: The knowledge team regularly met, often daily, to share progress, obstacles, and insights. Similar to a software scrum, these stand-ups kept everyone aligned and identified issues early. For example, if a subject expert was busy with other work, the team could reassign tasks or adjust scope quickly. They also used simple Kanban boards to track article statuses like drafting, reviewing, and publishing, ensuring nothing was overlooked.

• User-Centric Design: Content developed with the end-user in mind (in this case, the support agents and customers). They applied user-centric design principles, such as consistent templates, easy navigation, and simple language, ensuring that articles were not only technically correct but also usable. Agents tested early drafts during real support calls, and any confusion or complexity was noted as an area for improvement.

• Frequent Review & Retrospectives: After each sprint, the team held retrospectives to evaluate what went well and what could be improved. These retrospectives weren’t just about the team’s process; they also reviewed knowledge usage analytics. For instance, if specific new articles weren’t being accessed or were not reducing call volume, the team dug in to understand why. Perhaps the article lacked proper metadata, the title was off-putting, or an indexing glitch failed to present it when the context required it. This practice of continuous learning ensured the KM initiative continued to refine itself. They also celebrated quick wins, reinforcing the value of the effort.

• Continuous Engagement: Crucially, the organization encouraged continuous feedback from the broader user base of the knowledge system. Support agents were prompted to flag incorrect or outdated content and suggest new article needs on the fly. Rather than treat the knowledge base as a library managed by a separate team, it became a shared, collaborative space. Frontline employees felt a sense of ownership, seeing their suggestions rapidly turned into improvements.

The result of this agile approach was not just a successful knowledge base launch, but a sustained culture of knowledge improvement. Over time, agents grew confident that the knowledge base was the best place to look first for answers (because it truly reflected the latest information). Building knowledge management systems that people want to use improves productivity, increases customer satisfaction, and helps create a strong culture of continuous improvement (not by talking about culture, but by implementing practices that result in a continuous improvement culture).

By freeing up time and reducing rework, agents spend less effort rediscovering solutions because the collective knowledge is readily available and continuously updated. Agile in KM creates a virtuous cycle where solving one ticket today makes it easier to solve a similar ticket tomorrow.

The AI Advantage: More Precise Knowledge at Scale

If agile methods provide the process for rapid knowledge capture and improvement, AI can turbocharge those processes. Over the past couple of years, we’ve seen an explosion of AI capabilities (especially generative AI and large language models) that directly address some of KM’s toughest challenges.

AI can accelerate and automate knowledge-intensive tasks. Here are a few practical ways AI and machine learning are supercharging agile knowledge management:

• Instant First Drafts: One big hurdle in KM is the time it takes experts to document their knowledge. Busy professionals may know the solution to a problem, but accurately documenting it is another task on their plate. Generative AI can help by producing a first draft of a knowledge article or answer, based on context and past data. An AI, trained on past tickets and existing articles, can automatically generate a draft knowledge article for the solution, which the engineer can then review and refine.
  
  Editing or tweaking a draft is much faster than writing from scratch, lowering the barrier for capturing new knowledge. Even new employees can ask an internal AI assistant common questions and get a draft answer drawn from prior company knowledge, which they can then verify. Effectively, AI plays the role of a seasoned mentor, offering a starting point so that no one has to wait weeks for an expert to write it down manually. This practice facilitates the efficient flow of knowledge and aligns with an agile cadence.

• Dynamic Updates and Continuous Learning: In a truly agile knowledge environment, content isn’t static; it should evolve as it is used. AI systems can monitor how knowledge is used and proactively suggest updates. For example, machine learning can detect when particular knowledge base articles are becoming “stale” (perhaps users frequently ask follow-up questions indicating gaps, or an emerging issue isn’t covered at all). Rather than relying solely on a scheduled review cycle (which experts often bypass or delay), AI can flag outdated information the moment it becomes irrelevant. With the right integrations, it can even update a Kanban board to prioritize and make its findings visible.
  
  Some advanced knowledge platforms use AI to automatically scan for new information (say, changes in a product spec, or a newly discovered workaround in a developer forum) and will notify the team or even update relevant articles accordingly. In essence, AI helps the knowledge base “learn” continuously, so the content stays current between human-driven update cycles. This pairs well with agile retrospectives, as the team receives concrete data on what content needs attention next, thereby removing much of the guesswork from backlog prioritization.

• Semantic Search and Q&A: Ever typed keywords into an intranet search and gotten 100 useless results? AI fixes that. Modern KM systems incorporate semantic search and knowledge graphs that better represent the underlying data, allowing answers to queries to be more precise.
  
  This means employees and customers can use plain language questions, and the AI will interpret intent, review relevant knowledge sources, and present a helpful result, even if the question didn’t exactly match the document text.
  
  AI-powered search increases the uptake of self-service knowledge (people won’t use what they can’t easily find). It also provides feedback: if users are asking questions that yield no answers, that highlights an area where new knowledge is needed (which goes into the backlog).

• Summarization and Transformation: Another boon from AI (especially large language models) is the ability to transform existing knowledge into the format or detail level needed, on the fly. For instance, a company might have a long technical policy or a 50-page product manual. Previously, a human would have to manually write a concise “how-to” article or a troubleshooting guide from those sources, which is a time-consuming process.
  
  Now, an AI can be prompted to summarize the policy into a 5-step instruction or extract the troubleshooting segment from the product manual and format it as a Q&A article. I’ve seen organizations feed raw documentation into AI and get immediate draft knowledge articles tailored for different audiences (e.g., a “for technicians” version and a “for end-users” version of the same info).
  
  This dramatically cuts down the time between knowledge existing somewhere and knowledge being usable for decision-making or support. Every knowledge manager knows there’s a wealth of untapped information floating around (in white papers, in experts’ heads, in email threads). AI acts as a converter to make that knowledge accessible in the formats people need on a day-to-day basis.

• Augmenting Human Expertise with Broad Knowledge: Agile teams focus on leveraging internal knowledge, but sometimes an organization hasn’t encountered a particular problem before. Here, AI’s ability to synthesize external knowledge becomes valuable. Suppose an internal knowledge base is thin on a new technology that a company has just started using. Train an AI model on both internal content and relevant public information (such as open-source documentation, forums, and industry research).
  
  When an employee asks a question, the AI can draw from both sources, providing a blended answer that combines internal best practices with external insights. This ensures that even a smaller firm without decades of internal documentation can still offer helpful answers by learning from the outside world. It’s like instantly scaling the knowledge base with the world’s information, curated to context. (Note the suggestion of training, even on knowledge that the organization hasn’t created itself.
  
  [Note: If this step is skipped, an LLM, presented with a question for which it hasn’t been trained, is likely to make up an answer, often in a convicing and vigorous way. In addition, system prompts and other safeguards should be in place that ensure if an LLM doesn’t know something, it doesn’t hallucinate an answer.]

• Conversational Interfaces and Self-Service: A persistent challenge in KM is getting end-users (employees or customers) to actually use the self-service knowledge available. One reason self-service sometimes disappoints is the experience – users might have to navigate separate windows, small fields on a customer record, or read through lengthy articles. AI answers delivered via conversational interfaces (chatbots or voice assistants) provide knowledge in a more interactive, human-friendly way.
  
  Instead of searching a portal, an employee can ask the chatbot, “How do I submit an expense report in our new system?” and get a tailored, concise answer or even a step-by-step dialog. This is a game-changer for adoption. It turns knowledge access into a Q&A dialogue, which is often more intuitive and user-friendly. Conversational interfaces embrace the “power of pull” idea, as fostered by John Seely Brown and his collaborators, making it more likely that users will prefer self-service because it feels easier and more personalized.
  
  In agile terms, it’s about improving the end-user feedback loop: if users trust and like the knowledge tool, they’ll use it more, which in turn deflects more routine queries away from support teams and gives those teams more bandwidth to improve content, a positive cycle.

• Predictive Insights and Gaps Detection: Going beyond responding to current questions, AI can also analyze patterns in queries and content usage to anticipate knowledge needs. For example:
  
  AI might detect that several employees are searching for information about a new product feature that hasn’t been documented yet, flagging a content gap for the KM team to address. Alternatively, it is worth noting that one department frequently requests policy clarifications, suggesting that the official policy document may be unclear and in need of an update.
  
  Some AI tools provide dashboards of such insights, effectively acting as an early warning system that proactively drafts or refines knowledge before an issue becomes critical. This predictive element aligns with the agile ethos of being proactive and adaptive; organizations aren’t just reacting to knowledge demands, they’re staying ahead of them.

AI is never a silver bullet – it works best in tandem with a strong knowledge culture (and quality data). Feed an AI incorrect or obsolete information, and it will happily regurgitate it with a confident tone. That’s why agile KM and AI complement each other: agile practices ensure feedback loops and human oversight to keep knowledge accurate, while AI adds speed and scale.

A human-in-the-loop approach is vital. Have experts review AI-generated content, and let users rate or comment on AI-provided answers. As I often remind clients, they should test their AI outputs and build in guardrails. The goal is trustworthy agility in knowledge management. With proper governance (for example, documenting how and when an AI suggests an answer and providing users with a way to verify sources), organizations can avoid the trap of “coherent nonsense” that sometimes plagues AI, and instead reap the benefits safely.

Knowledge Management for the Modern Age

Agile methods and AI are not just buzzwords; together, they form a powerful one-two punch to modernize knowledge management in corporations. Agile keeps knowledge work human-centered and iterative, ensuring organizations focus on the highest-value information and learn from each update. AI provides the automation and augmentation that makes managing vast amounts of knowledge feasible by handling the grunt work, spotting patterns, and scaling insights across the organization in seconds.

The end game is a KM function that is fast, responsive, and deeply integrated into everyday work. Picture this: a salesperson in the field asks their phone’s voice assistant for the latest product specs. The assistant pulls from an AI-updated knowledge base and answers in a friendly voice on the spot.

Meanwhile, the product support community at HQ receives a dashboard alert that a dozen people have asked about a specific pricing policy this week. They realize an update or clarification is needed. By Friday, they’ve published a tweak to that policy document.

In customer support, a new issue arises, and the agent finds an AI-suggested solution article drafted from similar past cases, which they refine and adapt. By the end of the day, the article will be available for all agents and customers. That is agile, AI-driven knowledge management in action: knowledge as a living, evolving asset that actively drives efficiency and innovation.

Regardless of the industry, the combination of agile practices and AI capabilities can liberate institutional knowledge from dusty archives (yes, even virtually dusty digital archives) and put it to work. It enables organizations to learn more quickly, respond more effectively, and continually improve their performance.

The companies already doing this are seeing employees make better decisions with up-to-date information, teams avoid “reinventing the wheel” because the wheel’s design is already in SharePoint, and customers get answers in seconds via chatbots that would have taken hours via email. They’re turning knowledge management from a bureaucratic chore into a strategic advantage.

As someone who has been watching this space for years, the convergence of agile and AI is fascinating. It’s fulfilling the old promise of KM: getting the proper knowledge to the right people at the right time, as AI has finally delivered an approach that aligns with the speed and flexibility today’s businesses demand. With the rise of AI, all organizations should revisit their KM strategies.

Apply agile principles to knowledge curation and management. Experiment with AI tools that can lighten the load and reveal new insights. Start small, iterate, and scale up. An organization’s collective knowledge is one of its most valuable assets, and it grows in value the more it’s used. By combining agile methods with AI, organizations ensure that this asset is continually sharpened, accessible, and actionable.

Winners in the for-profit, not-for-profit and the public sectors will be those who learn and adapt the fastest. Agile + AI-driven knowledge management is how to ensure that organizations not only keep up with change but thrive on it, leveraging knowledge as a renewable engine of efficiency, innovation, and insight.

Surveys, even when well-designed, only reveal what respondents believe to be true. They don’t capture what participants don’t know, haven’t experienced, or can’t yet imagine. That’s why survey findings are best treated as guideposts, not verdicts—signals of where organizations feel confident, where they’re uncertain, and where blind spots may be quietly shaping their futures.

The APQC Communities of Practice in Modern Organizations survey offers exactly that kind of snapshot: useful, but incomplete. Its greatest value lies not in the charts, but in the conversations and actions it should trigger. And right now, those conversations should be urgent.

We are at a moment when Communities of Practice (CoPs) should be thriving. Economic uncertainty, political instability, and the accelerating integration of AI into daily work demand structures that connect people, transfer know-how, and create shared agency in shaping the future of their disciplines. Done well, CoPs turn tacit expertise into collective capability, aligning that knowledge with business strategy and performance.

Instead, the survey results reveal a pattern: too many CoPs operate without clear design, measurable outcomes, or the authority to influence practice and policy.

I wrote a slide-by-slide analysis at SeriousInsights.net. A summary of observations and recommendations follows.

From Social Space to Strategic Asset

Many CoPs in the survey lack strong integration with business strategy. Too often, they are positioned as helpful “extra” spaces for sharing rather than embedded elements of organizational performance. This is a mistake.

CoPs should be treated as complementary organizational structures, with responsibilities that extend beyond networking or resource sharing. They should carry accountability for shaping policy, improving practice, and driving measurable outcomes. That requires more than goodwill—it demands design.

Design for Purpose and Measurable Value

Every CoP should start with a clearly defined business objective, not just a vague “priority.” That objective should be paired with two Objective and Key Results (OKRs): one for learning and one for performance. Quarterly “value stories” can connect those OKRs to real outcomes, making alignment tangible.

Leaders must also define sunset criteria. A CoP that hasn’t produced tangible outputs or seen healthy participation for two quarters should be restructured or retired. This discipline prevents stagnant communities from draining resources and attention.

Rethink Roles and Skills

The survey shows inconsistencies in the roles supporting CoPs. While facilitators, subject matter experts, and core members are common, essential functions like outcome stewardship and knowledge pattern editing are missing in many organizations.

An Outcome Steward ensures that discussions translate into action, while a Pattern Editor curates and refines contributions into reusable assets—roles that become even more important when AI is part of the workflow.

Leadership skills should also evolve. Facilitation and passion matter, but CoPs that deliver sustained value are run like products. Leaders should be trained in backlog grooming, experiment design, and metrics literacy, enabling them to adapt based on evidence, not just enthusiasm.

Make Participation Real Work

Too many organizations “encourage” participation without designing for it. Culture doesn’t shift by promotion—it shifts through policy and practice. If people don’t have time to participate, they won’t.

Managers should replace low-value meetings with CoP engagement, making it clear that community participation is legitimate work. Job descriptions and performance reviews should reflect that reality. Without structural reinforcement, CoPs risk being seen as extracurricular rather than essential.

Measure Impact, Not Just Activity

The survey highlights a reliance on basic activity metrics—attendance, number of posts, page views—that say little about value.

Organizations should focus on measures like:

• Cost savings from adopted practices

• Revenue increases or improved win rates

• Risk avoidance and compliance improvements

• Brand equity gains through thought leadership

These metrics tie CoP success to tangible business outcomes, making the case for continued or increased investment.

Integrate AI—Thoughtfully and Urgently

Any CoP not experimenting with AI at this point is missing an opportunity. AI doesn’t require heavy infrastructure to be useful—simply bringing an AI assistant into discussions is an implementation step.

Deeper applications like retrieval-augmented generation (RAG) assistants, pattern recognition, auto-tagging, and custom chatbots can amplify knowledge capture and reuse. But AI must be deployed with guardrails: clear sourcing, prompt standards, and regular testing to ensure trustworthiness. All great topics for AI’s need for KM CoP. See my 2025 APQC talk page for more details on how to apply KM to AI.

The Risk of Irrelevance

Without significant reconsideration, many CoPs will become even less relevant as change accelerates. In fast-moving environments, optional, underpowered communities will be bypassed in favor of faster, more authoritative mechanisms for knowledge exchange.

CoPs should be drivers of adaptation, not victims of rapid change—actively shaping how organizations respond to disruption, absorb new technologies, and sustain the human connections that make innovation possible.

Bottom Line

This survey is not a verdict—it’s a challenge. It calls for leaders to redesign CoPs with intention, integrate them into business strategy, measure their contributions in terms executives understand, and give them the resources and authority to deliver.

If organizations fail to meet that challenge, CoPs will continue to drift toward the margins. If they succeed, CoPs can become one of the most powerful tools for navigating uncertainty, harnessing collective expertise, and turning change into opportunity.

If AI initiatives feel unmoored—prompts living in shared docs, agents behaving unpredictably, governance fixated on principles more than practice—this piece from our core Serious Insights KM research offers a clear path forward. It reframes AI governance through a knowledge management lens, grounding decisions in the types of knowledge AI actually touches and transforms.

A Practical AI Knowledge Governance Framework: Mapping AI Approaches to Knowledge Types

The core argument

• We don’t need new AI taxonomies. We need to apply well-vetted KM distinctions to AI work.

• AI systems interact with different kinds of knowledge in different ways; when organizations ignore those distinctions, they build opaque, brittle systems.

• Treat knowledge types as anchors for governance—so policies, audits, and design choices map to how knowledge is stored, used, and changed in AI systems.

The seven knowledge types and why they matter

• Explicit: Codified assets like prompts, model metadata, and config files. Governance focus: versioning, documentation, and change control.

• Implicit: Inferred behaviors such as guardrails or agent tendencies. Governance focus: testing, behavioral benchmarks, drift detection.

• Tacit: Human know‑how—prompt craft, agent orchestration instincts. Governance focus: capture practices, communities of practice.

• Declarative: Facts and relationships (RAG sources, knowledge graphs). Governance focus: source verification, refresh lifecycle, confidence scoring

• Embedded: Knowledge in architectures and constraints (agent design, platform guardrails, token limits). Governance focus: design transparency, bias review.

• Procedural: Multi‑step “how‑to” sequences that agents execute. Governance focus: process mapping, risk flags, override/rollback.

• Contextual: Meaning shaped by situation, role, or time (dynamic retrieval, conditional guardrails). Governance focus: context logging, trigger management, audits for context leakage.

What this unlocks for governance

• Moves from abstract ideals (transparency, fairness) to operational scaffolding tied to actual AI artifacts and behaviors.

• Makes audits feasible: you can track prompts, configs, and metadata; benchmark implicit behavior; and document where context enters and shifts.

• Reduces failure modes like stale RAG corpora, undocumented guardrail changes, or invisible architectural constraints that shape outcomes.

A practical checklist for KM leaders

• Embed KM in AI product teams and pilots to connect intent with implementation.

• Audit current AI systems against the seven knowledge types—identify what’s governed vs. assumed.

• Stand up prompt libraries with annotations, access logs, and version control.

• Create test harnesses for agent behavior; define and monitor drift and anomalies.

• Establish lifecycle governance for declarative sources (RAG/graphs) with scheduled updates.

• Document agent architectures and platform constraints; review for fairness and unintended exclusion.

• Map automated workflows end‑to‑end with risk levels per step, plus override and rollback paths.

• Log context signals and define explicit context‑switch triggers to avoid leakage or misuse.

The key takeaway

Bringing KM back to the center of AI transforms governance from a compliance burden into a strategic advantage. By aligning AI artifacts and behaviors to well-known knowledge types, organizations gain traceability, resiliency, and clarity about what their systems know, how they know it, and what happens when conditions change.

Read the full framework and recommended actions on Serious Insights:

A Practical AI Knowledge Governance Framework: Mapping AI Approaches to Knowledge Types

Knowledge curation has always required clarity, discipline, and persistence. AI brings speed, pattern recognition, and reach—but only if it’s paired with well-designed human practices. Here’s how to operationalize curation so that it produces visible, ongoing value.

1. Define Knowledge and Align with Strategic Objectives

• Run a definition workshop: Bring together representatives from strategy, operations, and frontline teams to define what counts as “knowledge” in your context—data, documents, process maps, lessons learned, expert profiles, etc. Capture examples in a shared reference guide.

• Prioritize with a relevance matrix: Create a two-axis grid—strategic impact vs. operational necessity. Place each knowledge asset category on the grid to identify what to curate first.

• Use AI to map coverage gaps: Feed AI a sample of strategic objectives and relevant operational documents. Use it to suggest missing topics or under-documented processes.

2. Design User-Centric Content for Relevance and Findability

• Interview users, not just managers: Ask how they search for information, what terms they use, and where they expect to find things.

• Develop a “findability checklist”: Before publishing, verify metadata accuracy, use of plain language, appropriate tagging, and cross-linking to related items.

• Implement AI-assisted search tuning: Use AI analytics to monitor search queries with poor click-through rates, then adjust metadata or create new content to meet those needs.

3. Establish Robust Content Quality and Governance

• Assign asset ownership: Every item in your knowledge base has an “owner” responsible for reviewing and updating it at least quarterly.

• Set review dates and automate reminders: Use workflow tools (AI-assisted where possible) to flag content approaching review deadlines.

• Score quality: Rate each asset on accuracy, completeness, clarity, and currency. Track scores over time and retire assets that consistently fail.

4. Balance Explicit Codification with Tacit Knowledge Leveraging

• Capture tacit knowledge through “micro-harvests”: After key meetings, record 5-minute summaries from subject matter experts, then transcribe and store them with metadata.

• Build expert locators: Maintain searchable profiles with areas of expertise, recent projects, and preferred contact methods.

• Use AI transcription + tagging: Let AI handle transcription, auto-tag recorded conversations and link them to related explicit knowledge assets.

5. Foster a Culture of Knowledge Sharing and Trust

• Integrate sharing into performance reviews: Track meaningful contributions (and effective reuse) to knowledge as part of annual evaluations.

• Offer micro-recognition: Post “Knowledge Shoutouts” on community homepages for valuable contributions. You can also offer other forms of recognition, such as gift cards or points toward experiences, including attendance at industry events.

• Run safe-to-share workshops: Practice handling sensitive or incomplete knowledge so people are comfortable contributing without fear of error.

6. Embed Knowledge Curation into Daily Workflows and Processes

• Add capture points to workflows: In project templates, include a “knowledge assets created” field that must be completed at closure.

• Link curation to change management: When a process changes, trigger an automated search for related documents and notify owners to update.

• AI nudges: Use AI to prompt users to save, tag, or update content when it detects high-value interactions (e.g., solving a problem in chat).

7. Leverage Technology

• Select tools after defining use cases: Example: “We need AI to summarize meeting recordings and extract action items,” rather than “We need the latest AI platform.”

• Integrate into existing tools: Add AI-powered search to SharePoint or Slack instead of introducing another standalone tool.

• Pilot, measure, scale: Run small experiments, measure impact on retrieval time or quality, and only then roll out more broadly.

8. Nurture Communities of Practice (CoPs) and Knowledge Intermediaries

• Give CoPs a purpose charter: Define who they serve, what types of problems they solve, and how often they meet. (See more on nurturing CoPs here).

• Assign knowledge stewards: These intermediaries help identify key insights from discussions, tag them, and link to related content in the knowledge base. They also support knowledge curation activities conducted by others.

• AI-assisted summaries: After each community session, AI produces a topic summary, recommended reading list, and related experts to contact.

9. Measure and Demonstrate Value Continuously

• Link metrics to business outcomes: Example: “Time to resolve customer issues dropped from 5 days to 3” rather than “We added 50 new documents.”

• Core metrics:

  • % of content up-to-date (define what “up-to-date” means, such as how often it was updated).

  • Average retrieval time

  • Active users as a percentage of the employee base

  • Cross-team reuse rates (tracked by asset IDs in different contexts)

• Report quarterly: Share simple, visual reports showing KM’s contribution to speed, quality, revenue generation and cost savings.

10. Embrace Continuous Adaptation and Learning

• Run quarterly knowledge audits: Identify emerging topics, under-used assets, and outdated content.

• Test and learn: Pilot new curation methods (e.g., AI auto-tagging) with a small group, collect feedback, refine, then expand.

• Encourage “after-action” learning: Immediately after major projects or events, capture what worked, what didn’t, and what should change—then feed it into the knowledge base.

• Monitor language shifts: Use AI to flag when users start searching for new terms or synonyms so that the taxonomy evolves alongside the business.

Our world is awash in data. Organisations keep adding digital channels, capturing more signals, yet decision-makers often still struggle to extract meaning. Research shows that the sheer volume of information affects an organisation’s ability to identify, organise and share knowledge. Generative artificial intelligence (AI) is emerging as an ally, supporting knowledge management (KM) processes and, by extension, decision-making. The challenge is to weave AI into KM practice without losing sight of the human experience that breathes life into knowledge.

Turning knowledge into action

Knowledge is never static. Creation begins in environments that foster inquiry and experimentation. People synthesize ideas when they have space to reflect, collaborate, and question assumptions. Capturing knowledge requires intentional design; it involves bridging the gap between tacit and explicit knowledge, documenting lessons learned and decisions so that experience does not fade away when people move on.

Sharing knowledge is inherently social and depends on trust; digital tools do not automatically promote open cultures. The value of knowledge lies in its application: it must be easy to access, relevant, and aligned with workflows. Tools, including AI assistants, are meaningful only when they serve people and processes. Too many unintegrated systems distract from the work.

Why decision support demands a knowledge strategy

Decision-making under uncertainty benefits from timely and contextual knowledge. When data is scattered and unmanaged, people waste time searching and risk making decisions based on incomplete information. Studies indicate that artificial intelligence and machine learning can help transform knowledge into action: AI-powered systems can automate categorisation, extract insights from unstructured data, and personalise access to critical information.

Generative AI advances this further by creating new content and summarising complex sources. AI, KM, and decision-making are, however, interconnected and mutually influence each other. Adoption requires a holistic approach that considers technology, organisational readiness, and the complexity of the problems being addressed.

Knowledge management is the critical bridge between data and decision-making. AI and machine learning enable real-time extraction, semantic search, and dynamic knowledge flows, turning static repositories into provisional, continuously updated narratives. However, this shift introduces governance challenges. When knowledge is continuously evolving, organisations must ensure their trustworthiness, address bias and provide context for AI-generated outputs. This is where KM disciplines, such as metadata, lifecycle management, and communities of practice, support AI rather than the other way around.

The triad: people, process and technology

Decision support sits at the intersection of human judgment and algorithmic augmentation. Sustainable KM aligns people, processes and technology.

• People – nurturing human insight. Knowledge lives in people’s experiences and relationships. Encouraging reflection, mentoring and communities of practice helps tacit knowledge surface and flow. Leaders must model sharing, reward collaboration, and make clear that contributions to shared repositories are valued. AI tools cannot substitute for trust; they can, however, surface patterns that prompt new questions and connect people to experts more quickly.

• Process – embedding knowledge into work. Effective KM is embedded in the way an organisation plans, learns and decides. Detect‑reflect‑affect loops transform lessons into new action. Conducting knowledge opportunity and gap analyses guides investment in training, hiring or acquisition. AI-enabled knowledge management reimagines this cycle as a dynamic feedback loop—data is mined in real time; semantic technologies enhance retrieval; dashboards and predictive models embed insights directly into workflows. Processes must also enforce quality and context: curated sources, documented decision rationales, and guardrails around models.

• Technology – enabling scale and speed. No single platform delivers knowledge. Instead, organisations should build technology architectures that support collaboration, search and curation. AI amplifies these capabilities. Semantic search and AI assistants interpret natural language questions and surface hidden connections. Automated categorisation and intelligent insight extraction reduce manual effort and keep knowledge current. Yet technology remains an enabler; success depends on alignment with culture and clear governance. Tools should be usable, interoperable and governed as knowledge assets.

Building resilient and intelligent organisations

When people, process and technology align, knowledge becomes a living system. AI accelerates insight, but KM ensures that insight is contextual, accountable and shared. That system supports more resilient organisations—ones that sense changes, learn from them and adapt. It encourages employees to make decisions quickly and responsibly by integrating relevant knowledge into their daily tools rather than forcing them to search across scattered repositories.

Actions to align KM with decision support

1. Establish model metadata standards. Define and enforce schemas for all AI models—version, training data, fine‑tuning details and intended use. Treat models as knowledge assets.

2. Catalog prompts and queries. Manage prompts, decision templates, and analytical queries as reusable artifacts with version control and documented performance. This practice helps reproduce insights and supports auditability.

3. Create guardrail governance frameworks. Document the intent behind each guardrail. Track changes, conduct behavioral audits, and educate users about constraints to ensure AI outputs remain transparent and aligned with policy.

4. Map context configurations to business goals. Review and document context window settings for language models. Ensure that retrieval‑augmented generation (RAG) sources and context sizes align with current knowledge priorities and adjust them as needs evolve.

5. Audit knowledge sources. Vet knowledge graphs, taxonomies and external data sources for accuracy, bias and recency. Automated extraction must be balanced with expert validation.

6. Implement model lifecycle management. Track usage patterns, test responses across versions, and archive decisions tied to specific models. This preserves institutional knowledge as tools evolve.

7. Communicate uncertainty and bias. Help AI systems disclose the limits of their knowledge. Train decision-makers to interpret generative output as a starting point, not an end state, and to question sources.

8. Facilitate cross‑functional collaboration. Embed KM professionals into AI development teams. Encourage data scientists, engineers, and business leaders to co-design metadata, guardrails, and governance.

9. Support communities of practice. Convene practitioners who test, refine and operationalise AI models, prompts and guardrails. Shared experience accelerates learning and improves alignment.

10. Monitor and respond to emerging gaps. Establish feedback loops to identify when models become outdated relative to reality and when new knowledge needs to be incorporated. Develop protocols to update sources and retrain models efficiently.

These actions move KM from a back-office function to a strategic partner in decision support. They recognise that AI’s power comes not from magic but from disciplined stewardship of knowledge. By treating AI and KM as intertwined, organisations can turn information overload into insight and make decisions that are both smarter and more human.

Organizations rarely fail at knowledge management because people don’t want to share their knowledge. They fail because the experience of sharing and applying knowledge is poorly designed. Accessing and contributing to repositories feels like chores, processes feel like obligations, and technology feels like surveillance.

In Management by Design I reframe this problem: knowledge management is not an IT initiative or a collection of policies. It is a designed experience that must balance policy and practice, technology, and space—the three foundations that support intentional management. When these elements align, they create knowledge ecosystems where people contribute naturally, learning accelerates, and strategy comes to life.

I avoid the abstraction of culture in Management by Design, explaining that policy and practice, technology, and space act as the levers that create culture. Design the experience correctly using those levers and a culture will develop. Abandon vigilance in that design, and the culture wanes.

Designing Policy and Practice: Creating Knowledge Behaviors by Design

Policy and practice set the rhythms of organizational life. Too often, KM policies are punitive or reactive—mandates to “document lessons learned” after a project closes, or complex approval hierarchies for contributing to a knowledge base.

Managing by design transforms these into intentional experiences:

• Policies become enablers, not obstacles. Instead of requiring rote documentation, policies encourage storytelling, reflection, and peer validation of knowledge contributions. They make practice formal and measureable.

• Practices are designed for flow. Weekly knowledge huddles, open peer-review sessions, and rotating “knowledge champions” replace bureaucratic capture activities. Practices are those activities that happen ahead of policy. The become embeded in behavior, and may actually be less effective if they become formal.

• Rhythm and motion are embedded in the cycle of work: detecting opportunities to share, reflecting on meaning, and affecting outcomes with applied insights. People must pay attention to the world around them, and contribute to learning and the applicaiton of learning.

By applying design tdehinking, organizations prototype these practices, test them with real teams, and refine them to ensure they motivate contribution rather than compliance. The experience feels natural because it aligns with the way people already work and learn.

Designing Space: Physical and Digital Environments That Invite Knowledge

In Management by Design, space isn’t just about walls and desks—it’s the environmental context where knowledge lives and moves, and where people work.

Physical space matters because tacit knowledge thrives in proximity and interaction. Organizations that design hubs for spontaneous exchange—from informal breakout areas to project war rooms—see more voluntary sharing than those who rely solely on digital forms.

Digital space however, is equally critical. A well-designed intranet or knowledge portal isn’t a repository; it’s a navigation experience that makes knowledge perceptible and accessible. Layering AI-driven recommendations, conversational interfaces, and visual dashboards can transform abstract content into actionable insight.

Space design must also acknowledge variety and emphasis: employees need different spaces—quiet for reflection, collaborative for ideation, and immersive for learning. This variety mirrors the mental motion of knowledge work, encouraging deeper and more continuous engagement.

Designing Technology: Tools as Orchestrators of Knowledge Flow

Technology provides the connective tissue of KM but only succeeds when designed as part of the overall experience.

• Equitability and flexibility ensure that tools work for everyone, across devices and contexts. Mobile-optimized knowledge portals, integration into workflow tools, and lightweight contribution mechanisms invite participation from all corners of the organization.

• Forgiveness and simplicity reduce friction. Version control, automated tagging, and AI-driven summaries help people contribute without fear of making mistakes or being buried in process.

• Adaptive intelligence—from search personalization to generative AI assistants—serves as a co-pilot, not a replacement. These tools detect patterns, suggest insights, and anticipate user needs, but the design ensures human judgment stays central to validation and action.

Technology alone never solves KM. But technology, when consciously designed in conjunction with appropriate policies, practices, and space designs, produces an environment where knowledge can move with rhythm and purpose.

The Experience of Knowledge by Design

When we bring these elements together—policy and practice, space, and technology—we design knowledge as an experience, not an obligation. Each foundation is orchestrated with the principles of perceptibility, proportion, and rhythm:

• Perceptibility ensures knowledge is visible and accessible in context.

• Proportion balances the demands of capture and use with the cognitive load on contributors.

• Rhythm and motion ensure that knowledge creation, reflection, and application become natural parts of the organizational heartbeat.

Design Thinking gives us the toolkit to empathize with knowledge workers, prototype new experiences, and iterate toward systems that feel fluid and human. Management by Design provides the architectural lens to align the organization’s foundations—policy, space, and technology—into a coherent ecosystem that adapts as people, markets, and tools evolve.

Organizations that approach KM this way move past static repositories and into dynamic knowledge ecosystems, where strategy and culture coalesce into an environment that learns faster, adapts faster, and innovates continuously.

Knowledge Management (KM) is a strategic discipline focused on leveraging an organization's collective intelligence to achieve business objectives. While technology provides the infrastructure for information flow, the true value of KM is unlocked when efforts are integrated with business strategy, cultivate a sharing culture, and empower employees to seamlessly access and contribute knowledge within their daily work. A critical element in fostering this environment is the implementation of effective reward and recognition programs, which address the fundamental human motivations for sharing knowledge.

The Imperative for Rewards and Recognition in KM

Knowledge fuels results, from innovative product designs to brilliant competitive moves. However, valuable knowledge, especially tacit knowledge, often resides within employees' minds and is difficult to articulate and transfer. Employees are often reluctant to invest their time and energy in knowledge capture and sharing activities unless they perceive an immediate and measurable benefit, answering the "what's in it for me?" question. Without proper incentives, tasks like contributing to a knowledge base can "fall through the cracks". Organizations also need to overcome cultural barriers to sharing, such as the "knowledge is power" mentality or the fear of being replaced if unique expertise is codified. Therefore, well-designed reward and recognition systems are essential to influence behavior, reinforce desired actions, and build a sustainable knowledge-sharing culture.

Categories of Reward and Recognition Programs

Reward and recognition programs in KM can be broadly categorized into financial/tangible incentives, formal recognition, informal/social recognition, and intrinsic/work-integrated benefits.

1. Financial/Tangible Incentives: These are direct or indirect monetary benefits or other valuable assets that motivate individuals to share knowledge.

• Direct Monetary Payments/Bonuses: Some organizations offer periodic bonuses to individuals for exemplary knowledge-sharing efforts. Hewlett-Packard, for example, established a "microeconomy” at one point, where contributors received monetary payments to their departments when others viewed or downloaded their material, ensuring quality through buyer-driven incentives.

• Linking Contributions to Compensation and Promotions: Making knowledge contribution a criterion for advancement or salary increases can be a powerful motivator. Consulting firms, for instance, have made it difficult for professionals to be promoted without significant contributions to knowledge-sharing efforts.

• Intellectual Capital Units (ICUs) and Stock Options: Points-based systems, such as "Silver ICUs" for primary knowledge behaviors (e.g., creating a profile, posting a question) and "Gold ICUs" for secondary behaviors (e.g., updating profiles, incrementalizing documents), can be linked to reward schemes, including stock options. This provides a complete audit trail for new knowledge ownership and appropriately rewards collaboration.

• Withholding Benefits from Non-Contributors: While a "stick" approach should be used sparingly, denying access to valuable information or resources to non-contributors can be a useful point of leverage. However, it's important to remember that knowledge seekers, especially junior members, are as crucial as contributors.

2. Formal Recognition Programs: These involve official acknowledgment from the organization, often in a public or documented manner.

• Public Recognition and Features: Contributions can be featured in company newspapers, through news desks, or via automatic publication of appreciation letters. Top management involvement in the recognition process, such as sending handwritten notes or ensuring visibility, significantly incentivizes contributions.

• Awards and Certificates: Organizations can offer specific awards for best practices. GTE Directories, for example, employs multiple broad-based recognition programs, often non-financial or with low monetary value, to reinforce business excellence. Clarica considered framed certificates, plaques, or symbolic awards for founding members of successful communities.

• "Halls of Fame" or Acknowledged Expert Status: Publicly identifying individuals as "company's acknowledged expert" or including them in a "hall of fame" can be a strong motivator, particularly for power-driven individuals.

• Linking to Performance Evaluations: Incorporating community contributions and knowledge-sharing behaviors into formal appraisal processes reinforces their legitimacy and importance. Rather than an extra activity, sharing knowledge becomes part of the job.

3. Informal/Social Recognition: These are less structured, peer-driven, or community-based forms of appreciation that foster a sense of belonging and mutual benefit.

• Peer Recognition and Gratitude: Appreciation from peers who understand the value of a contribution is highly effective. It is often more personal and informed than organizational recognition.

• Author Attribution: Ensuring that the name of the contributor is always associated with contributed knowledge in a system promotes ownership and visibility, allowing others to contact them for expert advice. Technicians, for instance, could be thanked and congratulated by peers at conferences for tips they've authored.

• Celebrations and Learning Forums: Events such as celebrations or learning forums can be important for individuals who value fun and foster a sense of community.

• "Knowledge Seekers' Club": Creating exclusive groups or clubs for knowledge seekers can appeal to individuals driven by love and belonging.

4. Intrinsic/Work-Integrated Benefits: These relate to direct improvements in an employee's work life or career progression as a result of KM participation.

• Increased Marketability and Expertise: Participating in KM activities, such as sharing knowledge, can improve an individual's marketability by increasing their knowledge and demonstrating expertise.

• Project Work Availability: Actively sharing knowledge can lead to opportunities for desirable project work.

• Professional Development and Learning: KM initiatives, especially through Communities of Practice (CoPs), foster continuous learning and skill development. People learn everything a system has to offer and go beyond it, continuing to learn.

• Enhanced Reputation and Visibility: Contributing high-quality and most-used content can boost an employee's reputation and make them a go-to expert within the organization.

• Access to Exclusive Knowledge/Communities: Participation in KM can grant access to valuable information sources or exclusive communities, leading to more effective job performance.

• Seamless Integration into Workflow: Embedding KM principles and actions directly into daily business processes makes knowledge capture and sharing organic, not an "add-on" or separate task. This minimizes the perceived effort of contribution. Providing employees with the time and space to contribute their best work is crucial.

General Observation about the Need for and Effectiveness of Rewards and Recognition:

Ultimately, the success of knowledge management depends on human behavior. While technology offers powerful tools for codification and transfer, it cannot create a culture of knowledge sharing; it can only enable and support it. The challenge is to motivate individuals to willingly contribute their knowledge, especially tacit expertise, which they may see as their primary source of personal value or leverage.

Technology and rewards can sometimes become a trap. I once worked on a lessons learned system that rewarded the “lesson learned of the week.” A lesson was submitted to a committee of managers via a database. The committee selected the winners.

I was brought in when contributions began to decline. In conversations with participants, I discovered that those who contributed but were never recognized decided to stop contributing. The underlying issue was the small pool of contributors and the equally limited group who could benefit from the lessons learned.

Our solution was to expand the system to all engineers, not just s small cadre in one business unit. Soon, while the formal system eventually wound down, the informal system of peer recognition drove more active participation.

By opening up the field, and offering a less one-way techology and making the reward peer recognition, the system flourished.

Engineers still received other recognition for writing up cost avoidance reports, but in this system, even seemingly minor suggestions often found someone who appreciated them.

Effective reward and recognition programs, whether financial, formal, social, or intrinsic, must be carefully designed to align with organizational goals and culture. They should focus on rewarding outcomes (e.g., highest-rated documents, documents linked to the most support cases, and the application of lessons learned) rather than just activities (e.g., the number of articles submitted) to ensure quality over quantity and prevent gaming the system.

Sustaining a KM environment requires ongoing re-evaluation of these incentives in response to environmental changes and feedback. When implemented thoughtfully, these programs reinforce the importance of learning, sharing, and collaboration, transforming knowledge management from a theoretical concept into a tangible driver of competitive advantage and enhanced organizational performance.

As organizations rush to digitize and automate using the latest AI techniques, they are quickly forced to admit that without a well-considered underlying structure, all that innovation ultimately rests on sand. That’s a primary reason it’s difficult to move from proof-of-concept with tight constraints, to large production systems that require access to much of the enterprise’s data, structure and unstructured. And that data may not be up to the task.

While knowledge management encompasses practices that require no technology, its ability to scale, to provide value to more than a few people, rests on an Information Architecture (IA) foundation. Those who are not seriously considering IA right now, especially as AI capabilities continue to accelerate, are falling behind.

IA Isn’t the Sizzle. It’s the Pan.

Information Architecture isn’t sexy, but it’s essential. It’s the classification system, the labeling scheme, the navigation plan. More than anything, it’s the cognitive map that allows people—and now machines—to find, understand, and act on knowledge.

If Knowledge Management is about making the right knowledge available to the right people at the right time, then IA is the mechanism that makes that possible. It doesn’t just organize; it translates chaos into clarity.

What IA Actually Does for Knowledge Management

Most people still associate IA with websites and intranets. But the discipline extends far beyond digital navigation. In a KM context, IA underpins everything from content discovery and collaborative workflows to semantic search and metadata standards. Done well, it doesn’t just reduce friction—it transforms the experience of interacting with knowledge.

Cutting Through the Infoglut

Most enterprises generate more content than they can manually manage. IA helps organize the information and make it meaningful. Through taxonomies and controlled vocabularies, content becomes navigable rather than just searchable. Without IA, even the best search engine is just a black hole with a pretty interface.

Taxonomies as Cognitive Infrastructure

Good taxonomies reflect how people think, not how content systems were architected. They serve as shared mental models that link related concepts, helping people move fluidly across topics. IA isn’t just about structure—it’s about semantics and context.

Metadata as Contextual Glue

Metadata helps make content findable, sortable, identifiable, and ultimately, usable. It acts as the connective tissue between disparate systems, topics, and use cases. But consistency is key. IA creates discipline for metadata so that it’s not just applied, but applied well.

From Content to Action

Information becomes knowledge when it’s situated in context and aligned with intent. IA doesn’t just help people find things; it helps them understand what they’ve found and decide what to do next. Contextual cues, relevant links, and embedded relationships guide users from passive discovery to meaningful engagement.

People-to-People Knowledge Flows

IA isn’t limited to documents and repositories. It also supports expertise location, enabling peer-to-peer knowledge transfer. Profiles, directories, skills matrices—all require thoughtful IA. Finding the right person should be as easy as finding the right document. AI is now creating new approaches, for expertise discovery, but they remain dependent on information consistency, even if they replace older, more manual approaches to directories.

Integrated Into Workflow, Not Bolted On

IA closes the gap by embedding knowledge access directly into systems and processes people already use. When IA is aligned with actual work, knowledge sharing stops being a chore and becomes second nature.

When AI Enters the Picture, IA Becomes Even More Critical

As GenAI, ML, and NLP reshape enterprise knowledge strategies, IA’s role expands from enabler to requirement. AI systems need structure, and IA provides it.

AI as the Great Organizer—But Only If You Help It

Yes, AI can tag content, summarize documents, and generate metadata. But that’s only useful if there’s a coherent schema behind it. AI can automate, but it can’t invent structure from scratch without guidance.

Knowledge Graphs Demand Discipline

AI thrives on relationships. Knowledge graphs offer semantic scaffolding for AI, mapping the connections between concepts, people, and data. But building effective knowledge graphs requires foundational IA work—clear entity definitions, shared vocabularies, consistent categorization.

Conversational Interfaces Still Need a Map

Just because someone can type a natural language query doesn’t mean the system knows where to go. Chat-based interfaces rely on underlying structures—indexes, ontologies, linked content—to generate coherent, contextual responses. IA is the backstage rigging for these systems.

AI Governance Starts with IA Governance

When AI summarizes, recommends, or cites content, accuracy matters. Strong IA helps enforce content governance—version control, data provenance, authorship, and usage rights. This is particularly important in regulated environments or when misinformation carries reputational or financial risk.

Without IA, Trust Suffers

People won’t trust AI if they can’t trace its logic. Explainability starts with structured knowledge inputs. If the AI pulls from poorly organized content with vague provenance, the result is uncertainty at best and hallucination at worst.

What’s Coming Next: Personal IA and Autonomous Agents

We’re entering a phase where users will no longer rely solely on enterprise-wide information architectures. AI agents—autonomous, purpose-built, and context-aware— will begin crafting personal information architectures tailored to individual workflows and preferences.

Imagine a system that observes how you interact with knowledge, the terms you use, the sources you trust, the decisions you make—and dynamically adjusts its metadata schemas and content organization to better serve you. This isn’t decades away. It’s the trajectory we’re on.

Agents won’t just be consumers of IA; they’ll be architects of it. And they’ll do it in real time, continuously re-structuring content ecosystems to optimize outcomes. That means organizations must build foundational IA that is modular, interoperable, and agent-friendly from the start.

These systems will also bring new challenges. Who owns a personal taxonomy? How do we reconcile personal and organizational ontologies? What does governance look like when machines generate metadata on the fly?

So Where Do We Go From Here?

Start with what you have. If IA has been neglected in your KM program, don’t treat it as a massive IT project. Start small. But start.

Here’s a pragmatic path forward:

• Audit what exists. Map your content repositories. Inventory your metadata. Review your taxonomies. Look for duplication, inconsistency, and disconnection.

• Talk to your users. Understand their workflows, frustrations, and habits. IA isn’t an abstract exercise—it’s grounded in behavior.

• Align IA to business value. Show how a better structure improves decision-making, accelerates onboarding, reduces compliance risk, or enables AI outcomes.

• Embed IA into governance. Integrate IA into your content lifecycle policies. Don’t bolt it on—bake it in.

• Design for adaptability. Agents will need flexible schemas. Build IA systems that can evolve—version-controlled vocabularies, customizable tags, and dynamic relationships.

Bottom Line

Information Architecture doesn’t get headlines, but it enables everything from content findability to AI-generated insight. It’s the infrastructure for intelligence. Ignore it, and your KM system becomes a landfill of forgotten assets. Invest in it, and you create a platform for clarity, creativity, and competitive advantage.

And soon, your agents will be building architectures of their own. Make sure you’ve taught them well.

Illustration from ChatGPT via a prompt written by the author.

Key Insights

• KM is not infrastructure—it’s experience architecture. It shapes how people interact with digital tools, not just what they find.

• People, process, and culture account for 80% of KM success. Technology alone will not bridge knowledge gaps or fix broken organizations.

• AI is reshaping the definition of knowledge. Generative systems challenge authorship, provenance, and the tacit/explicit divide.

• Infoglut is not just about volume—it’s about cognition. KM must actively reduce cognitive load and protect attention in high-pressure environments.

• Trust, transparency, and explainability are non-negotiable. Adaptive systems must be configurable and accountable to their users.

• Automation can’t replace discernment. KM systems need human oversight to prevent synthetic certainty from displacing nuanced judgment.

• Knowledge equity matters. Inclusive design must account for neurodiversity, cultural context, and varied cognitive approaches.

• KM must reconcile internal and external knowledge flows. Systems should curate, validate, and integrate open-source and AI-generated insights.

• Modern KM supports both operational excellence and strategic learning. It enables scenario planning, sensemaking, and narrative-based strategy.

• Investing in KM is a long-term resilience play. It fosters organizational memory, reduces risk, and unlocks value at every level of the enterprise.

In an era defined by rapid change, remote work, and an avalanche of information and misinformation, the traditional notions of work and how we interact with our digital tools are being fundamentally reshaped. It's no longer enough to simply have information; the critical challenge, and indeed the strategic opportunity, lies in how quickly and effectively we can leverage knowledge to drive innovation, make informed decisions, and enhance productivity. Knowledge Management (KM) should act as the silent architect behind a more intuitive user experience and a more efficient work environment.

KM: A Remedy for Broken (or Breaking) Organizations

While technology remains a critical component of KM, people, process, and culture, expressed through policy and practice, are paramount for its success. My previous work suggests that 80% of KM success comes from these non-technical factors. The recent explosion of remote and hybrid work models, massive layoffs, and even return-to-office polices, where interpersonal relationships often remain broken, has amplified the urgent need for robust KM, as organizations face increased knowledge gaps, information chaos, and the risk of losing critical expertise when employees leave.

Further, AI reshapes what organizations consider knowledge. It challenges authorship and provenance, and it alters the relationship between tacit and explicit knowledge. When knowledge informs the behavior of information systems, but users don’t understand the rules behind the behavior, those interactions risk becoming confusing rather than productive.

Knowledge Management's Footprint on Your Digital Experience

In the digital workplace, KM is the unseen force striving to make your virtual interactions smoother, more intelligent, and less frustrating.

1. Taming the "Infoglut" and Powering Findability One of the most immediate impacts of KM on user experience is the battle against “infoglut," the overwhelming abundance of information that can confuse and consume employee time without adding value. KM aims to deliver the "right knowledge, to the right people, at the right time”.

KM doesn’t just confront volume—it must also navigate human limitations. Cognitive load, attention scarcity, and relevance under pressure are equally urgent design considerations.

• Beyond Search: Traditional keyword searches often fall short, leading to "black holes of content". KM systems move beyond this by employing advanced search and retrieval mechanisms, often powered by Artificial Intelligence (AI) and Machine Learning (ML). These systems can automatically summarize documents, distil insights, and recommend relevant content based on a user's prior behavior and interests.

• The Power of Taxonomies and Metadata: To provide context and improve search accuracy, KM relies heavily on taxonomies and metadata. A well-designed taxonomy, ideally reflecting how users think about content, is crucial for organizing information and making it navigable. Knowledge graphs, first introduced in the 1970s, are now transforming search by semantically linking structured and unstructured data, revealing hidden connections, and enabling more robust and accurate queries beyond simple keywords.

• Intelligent Hubs: Modern intranets, now often called "employee experience platforms," are evolving beyond static content repositories to act as personalized portals. These aim to provide a "unifying interface" that streamlines access to disparate systems and content, making knowledge workers more productive. They are designed to deliver information that is "of immediate value to users". However, the success of portals hinges on robust content management processes that ensure quality, relevance, and context, coupled with collaborative features.

• Adaptive Workspaces and Generative AI (GenAI): The next wave of KM promises even more intelligent interfaces. Adaptive workspaces learn from user interactions, anticipating information needs and providing relevant tools and content without explicit requests. GenAI, such as Microsoft's Copilot, acts as a virtual KM assistant, simplifying complexity, mining existing knowledge, and even creating new content automatically (e.g., FAQs). This aims to free workers for higher-value tasks by automating "drudge work". However, careful governance and human oversight are crucial to prevent misinformation and ensure trustworthiness, requiring clear citations and mechanisms to limit "creative outputs" or "hallucinations".

• Equity and Accessibility: Though trends sometimes dilute the language of equity and inclusion, the imperatives behind it remain critical to KM. Systems must understand those who use the systems as intimately as it does the underlying data, leveraging the confluence of understanding to format, translate or represent knowledge in ways that make it accessible to the consumer. Is it an adaptive system, for instance, accessible to neurodivergent employees? Does the taxonomy account for diverse backgrounds?

The Limits of Automation in KM: Trust and Transparency

Adaptive workspaces powered by GenAI must evolve beyond delivering tailored content; they need to foster a participatory relationship with users. When systems suggest actions or surface insights, users must understand why the recommendation was made and retain the ability to challenge or refine it. Without transparency and control, adaptive features can become intrusive rather than empowering.

Trust in these systems emerges not just from accuracy, but from users seeing their intent and preferences reflected in the experience. Designing for explainability, configurability, and feedback loops transforms adaptive workspaces from reactive dashboards into co-creative environments where human and machine learn together.

Alternatively, those who support KM within enterprises must understand that those using the systems may place trust in external sources like Reddit or ChatGPT. Those systems have little or no knowledge of internal enterprise knowledge sources, and therefore may mislead and misinform through misplaced trust. That is an emerging KM issue that organizations must take on quickly to avoid safety and perception problems, and the liabilities associated with making poor choices that harm people or the brand.

Knowledge Management's Impact on Your Work Experience

Beyond the screen, KM deeply influences how employees work, learn, and interact with each other, fostering a more productive and fulfilling experience.

1. Cultivating a Culture of Sharing and Collaboration The most significant determinant of KM success is a corporate culture that actively encourages and rewards knowledge sharing. KM initiatives must acknowledge that knowledge sharing is a "highly voluntary" activity driven by internal choices, not just external stimuli or mandates.

• Motivation and Incentives: Employees are more likely to contribute if they understand the personal and organizational benefits—the "what's in it for me?". This can be fostered through recognition, performance reviews, and incentive programs, sometimes even tying KM contributions to stock options or "Intellectual Capital Units" (ICUs). See more on motivation in KM here.

• Facilitating Connections: KM emphasizes connecting people to people, not just people to documents. This is achieved through collaboration platforms, discussion forums, and the intentional nurturing of "Communities of Practice" (CoPs). CoPs serve as vital spaces for generating new knowledge and sharing tacit expertise.

• Knowledge Intermediaries: Roles like "knowledge stewards," "knowledge brokers," and "content managers" are crucial for identifying, capturing, and transferring organizational knowledge. These individuals, often embedded within business units, act as coaches and facilitators, helping people transform knowledge into usable forms and bridge knowledge gaps.

2. Embedding Knowledge into the Workflow: For KM to truly succeed, knowledge capture and sharing should be seamlessly integrated into daily business processes, rather than being perceived as a separate, time-consuming "add-on". This "in-the-flow-of-work" approach simplifies the KM experience and boosts adoption. Best practice organizations embed knowledge capture and reuse steps directly into methodologies like Six Sigma and Lean.

Modern knowledge work is porous—employees draw not only on internal systems but also on open-source platforms, peer networks, and generative AI tools. Traditional KM often neglects this flow.

Effective digital work design must reconcile organizational knowledge with credible external sources, creating a two-way channel that enables contextual validation, updates, and reinterpretation. This includes mechanisms for curating relevant public domain content, integrating APIs from trusted external databases, and creating policies that encourage safe exploration beyond the firewall.

The future of KM lies not in capture and containment but in orchestration, designing systems that make external learning a first-class citizen of internal knowledge work.

3. Fostering Continuous Learning and Innovation KM is intimately linked with organizational learning and innovation. It supports a continuous learning environment by providing access to past experiences ("lessons learned" and "best practices") and facilitating the creation of new knowledge. By supporting processes like "learning before doing," "learning while doing," and "learning after doing" (e.g., After Action Reviews), KM enables organizations to continuously improve and adapt.

• Empowering Knowledge Workers: KM aims to empower employees by providing them with the "right tools" and the opportunity to contribute their own expertise. It fosters a mindset where every employee is recognized as a "knowledge manager/knowledge worker".

The Limits of Automation in KM: Don’t Trade Insights for Speed

While automation can streamline routine tasks and enhance findability, it risks flattening knowledge into predefined templates that fail to capture nuance. Generative AI may produce plausible summaries or recommendations, but without context or human validation, those outputs can perpetuate outdated assumptions or introduce subtle errors.

KM systems that lean too heavily on automation risk displacing expert judgment with synthetic certainty—trading insight for speed. True knowledge work demands discernment, reflection, and negotiation, none of which can be reliably outsourced to algorithms. Effective KM design must include checkpoints for human interpretation, especially when decisions carry strategic, ethical, or cultural weight.

KM also serves as a risk buffer, ensuring compliance, preserving institutional memory, and making decision trails auditable. When done well, KM reduces the exposure created by turnover, misinformation, and reactive decision-making.

The Strategic Imperative of Knowledge Experience

Ultimately, knowledge management is about people learning and applying that learning. It connects the three core components of any organization: people, process, and technology, all wrapped in the social fabric of the organization.

While technology provides the infrastructure for information flow, the deeper value is unlocked when KM efforts are deeply integrated with business strategy, cultivate a sharing culture through policy and practice, and empower employees to access and contribute knowledge within their workflow seamlessly.

The most sophisticated KM systems have always supported strategy through functions like competitive intelligence and lobbying strategy. To a lesser degree, KM has supported scenario planning, futures thinking, or sensemaking in uncertainty, though it usually plays a much more operational role.

Because AI makes KM a better strategic partner by enabling organizations to challenge assumptions and nurture strategic learning, it elevates KM from an operational competitive advantage to a discipline that drives differentiation and shapes market action. My work on strategy focuses on strategy as an ever-evolving organizational narrative. My strategic planning approach requires KM by design.

Read more Serious Insights on Strategy

The success of KM is not a "silver bullet" or a "quick fix", but a long-term journey that requires continuous investment of time, energy, and resources. The impact can be significant, leading to improved communication, enhanced collaboration, better decision-making, increased employee skills, and improved productivity.

For senior leadership, understanding and investing in KM isn't just about efficiency; it's about building a resilient, innovative, and competitive organization where knowledge drives every strategic move. It's about designing an experience where the computer truly acts as an intelligent partner, and work becomes a continuous cycle of learning, sharing, and value creation.

Knowledge management has always been about human augmentation. Not in the science fiction sense of neural implants or cyborgs, but in the more grounded, yet equally radical, idea that people perform better when the right knowledge surrounds them—accessible, trusted, relevant, and primed for application.

What’s often missed in discussions about KM is that it isn’t about systems or repositories. Those are scaffolds. What matters is the scaffolding’s ability to help humans climb higher, faster, and more confidently. A robust KM program doesn’t simply preserve knowledge; it amplifies human performance.

Human Limitations Aren’t a Flaw—They’re a Design Parameter

We forget how bad people are at remembering things in detail over time. We confuse confidence with accuracy. And we habitually reinvent solutions, often unaware that an answer already exists just outside our cognitive periphery. KM intervenes in this cycle. Not by pretending to solve the human problem, but by complementing it.

Take onboarding. New employees don’t need a document dump. They need a curated path through ideas and experiences that are coherent and connected to their roles. KM at its best acts as an onboarding guide, not a file server. It curates, contextualizes, and calibrates knowledge so that new minds don’t get lost in a thicket of outdated PDFs and out-of-context Slack threads.

AI Doesn’t Make This Less Human—It Makes It Sharper

AI sharpens the edges of KM by doing what humans shouldn’t waste time on. Generative tools classify, summarize, translate, and extract. Predictive systems identify usage patterns and gaps. Semantic engines align terms across disciplines and departments. The machines aren’t thinking; they’re pattern amplifiers. And in KM, pattern recognition often equates to insight velocity.

Where I find AI most valuable is in surfacing the unknown knowns—those tucked-away pieces of insight that never made it to a formal repository but live on in chat logs, meeting transcripts, or even the rhythms of project management updates. AI can listen at scale in a way people never could, not to replace human understanding, but to elevate it.

The Augmentation Layer: KM as a Cognitive Companion

Effective KM doesn’t sit behind a search bar. It follows the work. It knows the difference between content and context. If you’re crafting a marketing plan, it should surface customer personas, past campaigns, market research, and team insights—not just because they exist, but because they are meaningfully entangled with the task at hand.

This is where AI has the potential to move KM beyond static libraries. We’re entering a phase where knowledge systems can become proactive. Not pushy Clippy-style interventions, but adaptive companions that understand timing, role, and relevance. For instance, an AI-powered KM layer might recognize that a project manager is planning a sprint in a regulated industry and automatically surface compliance constraints, prior project artifacts, and team member expertise profiles without a query ever being made.

That’s augmentation—human-centered, context-sensitive, and time-aware.

KM Doesn’t Eliminate Expertise—It Elevates It

A pervasive myth in KM circles is that good KM flattens hierarchies of knowledge. It doesn’t. It surfaces patterns, which is not the same thing as making everyone an expert. Expertise still matters. What KM does, especially when powered by AI, is reduce time spent rediscovering. It puts past work in conversation with present needs, allowing experts to spend less time searching and more time synthesizing and evolving.

Too often, knowledge workers are asked to perform triage on information: is it current? trustworthy? duplicative? relevant? With a KM system tuned for augmentation, those filters are applied before the person enters the equation. AI helps maintain the hygiene of the knowledge base so the human can focus on what machines still can’t do: interpret subtlety, nuance, and strategic implications.

Policy and Practice Still Arbitrate

Technology can scale, filter, and accelerate, but culture, expressed through policy and practice, enables people to act. If an organization isn’t ready to engage with knowledge as a utility, not just a product, no system or AI overlay will fix that. KM thrives where sharing is rewarded, where reflection is institutionalized, and where leadership sees knowledge as both a strategic asset and a social process.

AI can surface a brilliant pattern of insight-sharing across project teams. But if performance reviews still reward hoarding over openness, that signal will be ignored. Augmentation requires alignment between technical capability and cultural intent.

A Final Thought on AI and the Human Arc

There’s a temptation to see AI as the new knowledge worker. But that narrative skips over the core truth: humans are not being replaced, they’re being reframed. Augmentation isn’t about offloading what makes us human—it’s about shedding what holds us back. When AI takes on the repetitive, the extractive, the summarizable, it clears the stage for deeper insight, better storytelling, and more meaningful decision-making.

The real power of knowledge management in the age of AI isn’t in its novelty. It’s in the reassertion of what matters: people, empowered by what they know, able to learn from what they’ve done, and increasingly guided—not governed—by machines that can help them be more of what they’re already capable of becoming.