Work Trends OWL

Double-Loop Learning with AI: Getting Past the Why Barrier

Deepa Premkumar — Wed, 19 Aug 2026 03:31:23 GMT

We ended the post on values and identity with a simple looking instruction. When you want to tell a real value from a habit wearing its clothes, ask why. Why do I do this? Why does this task matter?

Asking ‘why’ sounds easy. It is not.

When Why Becomes a Question About You

Have you noticed what happens when someone asks you a why question? Why are you late? Why is it done this way? It pushes you on the back foot. A ‘why’ question makes us feel defensive. Like we have to explain ourselves. God help us, if we get the tone even slightly wrong. “Why did you do that?’ lands as what on earth were you thinking!

So, the trouble is, the answer to a ‘why’ question can be powerful and set us free. Yet ‘why’ is a question we would rather avoid.

Now think of the question from the previous post of identifying what we value with AI: “Why do we really do this task? Even before we realise, the conversation shifts from the behaviour to us (the person). And the moment we feel our competence, intent or credibility is being evaluated, we defend. We tend to protect ourselves rather than explore possibilities.

This is not stubbornness or a character flaw. It is human wiring, a protective reflex, doing its job. (Robert Kegan and Lisa Laskow Lahey).

So, we take a successful way out of answering a ‘why’ question.

Single Loop with AI: Looks a Lot Like Success

We can work with AI, without ever bringing up the ‘Why’ question. And it feels successful.

We continue with the same tasks in a faster manner and we never ask, Why do we really do this task? We use the latest frontier models. We use new AI tools. We produce more and faster. The report that took a full morning now takes ten minutes. Every box ticked. The output and calendar agree that we are doing well.

This is single-loop learning and it feels safe. Single-loop learning keeps the goal fixed and simply focusses on getting better at reaching it. The term Single-Loop and double-loop comes from Chris Argyris and Donald Schön. They described two very different ways of responding when something changes.

Think of a thermostat controlling temperature in a building. It would be set to a particular temperature - say 25 degrees C. When the room heats up, the thermostat notices and cooling comes on. When the room temperature goes below 25 degrees, the cooling switches off. Very reliable. Yet utterly incapable of asking the important ‘why’ question. “Why is the temperature set to 25 degrees? Is that the right temperature at all?

Most of us are using AI like that thermostat, in single loop mode. We have pointed a remarkable tool at the same old set of tasks and asked it to get it completed faster. We look successful. We are running in exactly the same place.

Double Loop with AI: Breaks The Why Barrier

There is a problem hiding in the thermostat, and I sat with it for a while. A thermostat cannot question its own setting. It has no way to stand outside itself and wonder whether twenty-five is the right temperature. Here is the uncomfortable part, neither can we, atleast not easily.

The instrument we would use to check our values is our values. You cannot spot your own comfort zone using the mind that built it. This is a blind spot in the truest sense. Not something we are too lazy to notice, but something we cannot see from where we are standing. Asking “am I only being comfortable here” from inside the comfort zone is like asking the thermostat to doubt itself.

Double-loop learning with AI is the response which not only closes the gap, but questions the ‘why’. Double loop does not ask, how do I write more code with AI? It explores the ‘why’ of writing code. Why was the code written? What did I learn by writing this code? Not just How do I do this task faster? but ‘Why should this task be done?

It does so, by separating the practice from the person. Reflecting on what you did is a non-defensive route to the why. Because you are examining the outcome and the practice, not putting yourself on trial. Double-loop asks what am I learning here? and not just did we achieve the target. By separating the individual from the task, the defensive guards come down. Finally learning gets through.

AI is that opportunity or jolt, which can trigger double-loop learning.

The Loop Runs Both Ways

The typical loop of values, actions and identity runs like this. What we value drives our actions, and our actions harden into who we are. Value, then action, then identity.

This series of our identity with AI, argues the opposite. AI is providing us an opportunity to change our actions first. And consistent new actions teach us to value differently. Action, then value, then identity. Both are true. This is not a contradiction but a loop that runs both ways.

Most of the time we enter through the value door. We do what we believe matters. But when the value itself is unclear, and asking the ‘why’ only gets our defences up, we cannot get in that way. So double-loop learning addresses the ‘why’ question through the other door. Through action and reflection. AI provides the trigger for this reflection.

I could never question why my Friday report existed while I was busy writing it every Friday. Now AI writes it in four minutes. It frees up too much bandwidth to ignore and pretend like nothing has changed. AI is allowing us to ask the ‘Why’ question.

AI & The Why Question

You do not try to win the argument with your comfort zone. You run a small experiment instead. You let AI take the Friday report, and you spend the reclaimed hour on the one conversation you have been avoiding. Nothing is decided. Something is tried. And if it turns out to matter, you notice your sense of what is valuable has quietly shifted.

This is why we are not thermostats, and it is the whole point. A thermostat can only hold the setting it was handed. We can question ours, not by staring harder at the dial from inside the room, but by letting the jolt, and then our own small experiments, reset it. Each experiment is a vote for a different value. Cast enough of them, and the role you hold is no longer the one your task list handed you. It is one you have recrafted.

That practice: acting, and then talking and reflecting your way into who you are becoming, is where this series goes next.

Value or Habit? What AI Is Forcing Us to Ask

Deepa Premkumar — Tue, 04 Aug 2026 09:05:42 GMT

For five years I headed a committee. It started off because i wanted to contribute. Soon the impact and applause followed. I enjoyed the visibility and feeling of being good at it. I over stayed my role at the committee. I continued without realising the applause was driving me instead of the impact. Worse, i was using the committee as an excuse of busynness rather than go after other outcomes, that were scarier but crucial for my growth.

It took me a long time and courage to realise and admit that, even to myself. What i was calling and valuing as commitment was simply staying in my comfort zone.

Values are tricky. What we value, tends to lead us towards aligned actions. Very soon these repeated actions and values become our identity. One hard to separate from the other.

Earlier in this series I argued that we can build new identities by trying things, by taking actions. Identities need not always be pre-decided or unchangeable. But you don’t try things or take action at random. Some trigger or reflection has to prompt you. Giving us a clue or next step towards what to reach for and what to leave behind.

The trouble is, life keeps us busy. With our noses at the grindstone. Never ending to-do lists and deliverables. When priorities beckon, quiet spaces for reflection are the first to go. Triggers are ignored and we keep busy. We don’t look closely and we continue to run on things that we call our values.

Value or Habit?

Sometimes it is a habit that has been with us for long, it becomes what we value. How many times have we not heard the phrase - this is how do things here! Think of a report that goes out to a distribution list every friday. Ask why, and the honest answer is that it has always gone out on Friday. That is not a value. That is inertia or habit in a better outfit.

Value or past success?

Sometimes it is an old success that has hardened into a rule. A move that worked brilliantly once. A playbook that saved the day years ago. Ask why, and you realise it worked so well that we stopped asking whether it still fits.

Value or comfort zone?

And sometimes, as I eventually admitted about my committee, what we are protecting is comfort. We know how the role or task works. Staying gives us the warm feeling of competence, and we dress that feeling up as our identity. It is the sneakiest of the three.

Asking ‘why’ we are doing something is the best start point in separating a habit, a past success or a comfort zone from our identity.

AI Forces Us To Ask ‘Why’!

And when AI absorbs so many of our tasks, it is a window an opportunity to pause and ask, “why”? Why did we do these tasks in the first place? What value did they bring for us?

We have been too busy, to ask the question of ‘Why’ and we have been getting away with it. Now, with AI starting to free up our time, we have an opportunity to clarify and craft the ‘Why’. A small window now, before it is settled for us.

This is an opportunity, for sure. Who we are at work was decided by the tasks that filled our day. Now, we get to craft it and answer who we are, with more deliberation. A chance to ask, ‘why’ are we doing ‘a’ particular task. Should it be delegated/automated/removed. But answering the why question is not as easy as it looks!

Noticing Our Emotions With AI

When you sit with the why, watch your own reactions closely. Notice which tasks you rush to defend, and how quickly the defence arrives. Notice too, where letting go feels more a sense of loss. That feeling is worth naming because grief is sneaky. It often shows up as a complaint about AI’s output, when what we are really mourning is a task which had become a part of us.

Notice what you are valuing through the tasks you complete

Both columns matter. Neither is wrong. But the left tends to keep us in the same orbit and the right compounds and lifts us up.

Valuing capacity building with AI

What you value will define how you use AI

If you continue to value the habits, the comfort zone, the dopamine hit of task completion, you will resist AI. The resistance will show up in small, passive and often not easy to spot ways. If you begin valuing AI as a capacity builder, you possibly will play with AI differently. Valuing the possibility of doing what was not possible before. Rather than merely doing the old things a little faster.

Research over 100 years have shown us, we build identity and through what we value and small consistent actions. Each time you use AI to reach past your strength, instead of defending your ground, you become someone a little different. Someone whose worth is no longer tied to the tasks AI has taken. But defined by the capacity and capability being built.

This is the most energising possibility with AI. Not a faster version of who you were. You are not just choosing how to use a tool. You are crafting what you value.

A brief window, open now, to choose who you become with AI. I missed a window like that on the committee. I do not plan to miss this one.

How Is Identity Built In The Age Of AI

Deepa Premkumar — Wed, 22 Jul 2026 08:04:15 GMT

An argument with a teenage daughter really makes you question your identity. In this case, the argument was about my AI usage and if I genuinely bought the expertise and was doing the work. In her telling, I had not gotten smarter. I had just found a smarter intern. I called it directing. She called it delegating to a machine and keeping the credit. She disagreed and as always walked off in a huff.

The Unease Professionals Are Carrying

Milder versions of this conversation are happening within us and at water cooler chatters. The unease we sense among professionals is what they really bring to the table. What is my identity in a world where AI is automating tasks rapidly?

Identity questions are not a sign that we are broken. It does not mean something is wrong and needs to be fixed. Identities shape and evolve over time. Unfortunately, sometimes much of the identity we carry was never really chosen by us. It somehow evolved. AI is giving us an opportunity to choose our identities deliberately. By automating tasks and compressing a large part of work, AI is freeing us up to ask the identity question.

Many a times identity evolves by default. Things we are good at, get adopted as our identities, without conscious thought. William James, an American philosopher, noted that we carry multiple identities. We move between them depending on who we are with. Erving Goffman, a Canadian-American sociologist showed us that we may carry an identity to display and another backstage version we keep for closer groups or ourselves.

Identity Is Curated, Not Delivered

But one thing which stands out from William James in 1890 to the identity researchers of today, is that we do not wake up to an identity one fine day! There is no voila moment. We grow into our identities, many a times, sub consciously.

Think of it as shopping for a new look at the store. You browse the shelves. You pick out a few items to try on at the fitting room. You see how it looks and feels on you. You mix and match outfits and accessories to get the look you are comfortable with. You choose what works best and leave the rest behind. This is closer to what recrafting an identity feels like. You are not accepting a ready-made version handed to you. You are curating one from the pieces that fit.

Herminia Ibarra, has spent her career studying how professionals reinvent themselves. Her findings match our fitting room experiences. You won’t know for sure which identity suits you best, until you try it on. Defining who we are in this AI driven world, will be no different. We will have to curate our identity through experiments and conscious choices with AI.

Experiment Your Way to AI

Think of you using AI as experiments. Each experiment is a different self we are trying on and not merely handing off tasks. The experiments stack up over time and give us fresh perspectives. With every new way of working we are curating our new identities.

I may have lost the argument with my daughter. But the unease i felt is good enough to make me ask myself if being an AI handler is enough. Made me question, how I can build something so different, that it justifies my AI addiction :)

So I will leave you where she left me.

Notice the unease. Do not rush to fix it. Stop waiting for the answer and your new identity to arrive before you move forward. It does not work that way. You experiment your way to it.

AI And The Case Of Mistaken Identity

Deepa Premkumar — Tue, 07 Jul 2026 07:30:03 GMT

Have you noticed typical introductions when you meet someone? “Hi, My name is ..so and so.... and i am ..role designation... at ..company”. Our role designations have got intertwined with who we are. But scratch that surface a bit and ask them what they do, and they will invariably respond with a set of tasks.

Many times, what we do is an easier question to answer than who we are. Organisations compound this with the way Job descriptions are written. JD’s are a set of tasks and not about impact. Yes, tasks give us a sense of accomplishment. And over the years, maybe without meaning to do so, we have come to define ourselves with the tasks we do!

Which is why AI automating tasks at such speed is a moment of reckoning and reflection.

The Question Nobody Warned You About

Picture a software engineer who has spent fifteen years defining herself through the code she wrote. Now, AI begins writing the code. She finds herself staring at a screen one afternoon, wondering something she has never wondered before. If I am not the one writing this code, what exactly am I here for?

This question is starting to get more and more and more common. It is uncomfortable and disorienting. But this is not a productivity or an execution focused question but rather a question of identity.

Why Tasks & Execution?

Tasks and Execution have a lot going for it. Tasks are urgent, visible, and satisfying. It produces something tangible. And in most organisations, the systems, the metrics, the performance conversations are built around outputs. As Cal Newport has observed, organisations tend to measure and reward what is easy to see. Tasks completed, emails answered. One thoughtful and hard conversation is harder to point to than a few smaller problems resolved.

Execution is familiar ground. It is where the professional identity was built.

This is where generative AI becomes genuinely interesting. Not just as a productivity tool, but as something that surfaces a deeper question about role and identity.

Generative AI is compressing the execution end of the spectrum. Reports, emails, data summaries, routine analysis — tasks that once took hours now take minutes. When that compression becomes large enough to notice, something shifts. Not just on your calendar. The question of what you are really here to do starts to surface.

That question is not a role design question. It is an identity question. And it is precisely because it is an identity question that many leaders, even when they recognise it, find it difficult to act on.

Not the Same Thing

A teacher whose lesson plans are drafted by AI is still the one who knows which student needs a push and which one needs room to breathe. A lawyer whose first-draft contracts are generated in seconds is still the one who knows which clause will actually matter in a dispute three years from now. The task was never the whole story. It was the part that was easiest to see.

The software engineer in this post reminds me of parents facing an empty nest! For decades, parenting was built around a list of never ending tasks. Packed lunches, school pick up, homework help, bedtime. Then kids grow up and the task list empties out. The parent feels lost and disoriented, much like the engineer staring at the screen. But eventually, most parents realise something. Tasks are gone, but being a parent was never only the task list. It was always the thing underneath it; the worry, the care, the way they show up. The role did not disappear when the task did. It was easier to see the tasks than the role, so the tasks got mistaken for it.

This is what we are up against right now. Not a loss of identity, but a case of mistaken identity. The task and the self, held so close together, for so long, that it gets hard to tell where one ends and the other begins. Ruben Hassid, an AI strategist, took a similar question to Demis Hassabis, the Nobel prize winner and CEO of Google DeepMind. His reflections pointed to the same things - maybe screens and tasks were never the whole point.

Tasks are not without value. They give us real accomplishment and a way to measure a day well spent. But being useful is not the same as being who you are. It never was.

The Reframe AI is helping us with

Here is the part I keep coming back to. Maybe AI is not taking something away from us. Maybe it is handing something back.

AI is forcing us to stop taking the shortcut of tasks to define ourselves and our identities. Once our identity is separated from the tasks we complete, we do not rebuild it by trying harder at the next task. We question ourselves in curious and exploratory ways. Where does my sense of value come from?

That is exactly where this series goes next. Not what happens when the task disappears. But how an identity gets built in the first place, and how deliberately we can go about building it again.

Why Is Generative AI Asking An Identity Question

Deepa Premkumar — Tue, 16 Jun 2026 14:50:03 GMT

I was talking to my financial advisor recently and he said something that stayed with me. Every industry is being affected by AI, he said. Usually change comes for some sectors at a time. This feels different. It is hitting every industry, every level, every role - the new hire and the CEO alike.

He is right. And what makes it different is not just the wide impact of the change. It is jolting us out of things we assumed were inherently human. We never imagined that a technology or tool we invented could take away so much of what we do! And do it faster, better and cheaper by a significant margin.

Every technology before this changed what we do. The excel replaced the paper ledger. Email replaced the type written memo. Each wave automated a task, and sped something up. Yet we remained untouched in some ways. We had to adapt, learn the new tool, and carried on. We updated our habits. We rarely had to update our sense of who we are.

Generative AI is asking something harder. It is not just changing what we do. It is asking us who we are in a world where it can do so much of what we did.

That is a very different question. A harder one.

Why does AI feel different?

So why is this moment different? Why is generative AI making us question our roles, our sense of who we are? Our identities?

One, AI speaks your language.

Every tool before this required humans to learn its logic. We learnt formulas to use excel. We learnt software languages to code. To use generative AI, we do not need to learn any new things. We bring our natural language, our context, our intent, and AI meets us there.

Strangely and almost counter-intuitively, when we do not have to learn any specific skill to use a technology, a different question opens up. What am I actually bringing to the table? Just like the London cab drivers when GPS arrived. Suddenly the thing that defined them was available to everyone. Ease made their edge invisible.

This question is not a capability question. It is an identity one.

Two, Speed of AI Creation

Think about a piece of work that used to take you two days. A report. A proposal. A first draft of something complex. Generative AI can produce a version of that in four minutes. This is not an incremental improvement. It is an entirely different league and the difference is hard to ignore.

When speed and output is no longer our advantage, the question shifts. What do I contribute to the outcome is the question staring at us. And this is not a productivity question. It is a personal one.

Three, It removes the skill excuse.

With earlier technologies, not knowing how to do something was a legitimate boundary. “I am not a designer” or “I am not a data analyst” were reasonable things to say. Those boundaries gave us clarity about what our role was and was not.

Generative AI quietly removes that defence. You do not need to be a designer to create a visual. You do not need to be an analyst to analyse data. When the technical barrier falls away, a different and more uncomfortable question takes its place. Not what can I do, but what is it that only I can bring?

The Real Question

AI touches how we see ourselves. It forces us to ask the question: What is it that only I can bring?

Not what can I do faster with AI. Not what tasks can I offload. But what is the thing — the judgment, the instinct, the human quality — that I bring to this work that the technology cannot replicate.

This question is an identity question. An identity question is not about what you can do. It is about who you are when the doing is taken care of. Most of us have never had to answer that at work before. And it does not have a quick answer.

In the last post we asked three questions to help start that process: what to let go, what to add, and what to protect. But identity recrafting goes deeper than three questions. Over the next few posts, we will explore what that deeper work actually looks like and what it asks of us.

Identity Recrafting in the Age of AI

Deepa Premkumar — Tue, 09 Jun 2026 02:14:10 GMT

The previous post on AI in particular and technology in general, forcing humans to adopt new identities got us a few interesting conversations. One of the leaders said “Factories and the printing press were threats right from the start. They were always going to replace humans and hence it was apparent that we had to step up differently. But AI is an enabler, a technology which is helping humans do better. Why would we need new identities in such a situation?”.

The need for change is easier to see when there is a threat. We have seen this time and again in our work. Individuals and organisations that have their back against a wall rally differently. When revenues are threatened, when survival is at stake, new ways of working and new identities are easier to adopt. More often than not however, the need for change is strongest when things seem to be going well.

Take a radiologist for example. Technology currently helps assess scans far better. AI identifies anomalies more accurately. On the face of it, things are going well. Work gets done faster and more accurately. This is usually when professionals surrender control to technology and at best do more of the same. It is at this precise moment when professionals across industries should be asking three questions. These are not just questions about your work. They are questions about who you are becoming in your role. Answering them consciously is how identity recrafting happens. In the small deliberate choices you make about what to hold and what to release.

What am I letting go?

When technology steps up, it makes sense to release the work it does better. A radiologist who insists on not using technology is clinging to habit dressed up as dedication. There is a phrase worth holding onto here: letting go by choice is evolution. Letting go by default is erosion. Consciously naming what we are releasing puts things in perspective.

What am I taking on that will add value?

You cannot stop at the first question. The space that technology creates is not meant to stay empty. Nor should it be filled with more volume of the same work. For a radiologist it means getting better at talking with patients, at seeing clinical connections across cases, at bringing judgment to the situations where AI flags something but cannot interpret what it means in context. Accountants face the same question — releasing the arithmetic to focus on the insight and advice that numbers alone cannot give. Architects who let CAD handle the drafting find themselves with space to do the thinking that software cannot replicate.

If this technology disappeared tomorrow, would I be inconvenienced or handicapped?

There is an important difference between the two. Inconvenienced means the tool is gone but the skill remains — you can still do the work, just more slowly. Handicapped means the capability itself has gone. Do not dismiss the question. It is playing out with commercial pilots right now. Pilots are so accustomed to automation that many hardly touch the controls. They have begun to lose the core skill they trained for. The Federal Aviation Administration has asked pilots to deliberately switch off autopilot so they can practice and stay sharp. Because if they need to override the controls in an emergency, they need to know how. That capability, once lost, does not come back quickly. It must be practiced deliberately, kept alive by choice, and treated as non-negotiable.

These three questions are not a one-time exercise. The tools will keep improving. Roles and identities will keep shifting in response. The professionals who do well are not the ones waiting for a crisis to force their hand. They are the ones asking these questions now, while everything still feels fine.

Recrafting your identity is not a crisis response. It is an ongoing practice.

And it can start with three questions.

AI is asking an Identity Question Of us

Deepa Premkumar — Tue, 02 Jun 2026 01:30:56 GMT

Everyone is talking about what AI can do and how quickly it is getting better. When the conversation moves further, it lands on job losses. Very few people are talking about what it does to us — our sense of who we are at work.

Every time a powerful tool has arrived and changed what is possible, the deepest disruption has never been to the work. It has been to the person doing it.

This has happened before. Not once, but repeatedly. And each time, the people who navigated it well asked a harder question: who am I now, and what do I value?

History has some lived experiences worth looking at, to understand how AI will impact our identities.

The GPS and the London Cab Driver

Let us start closer, in the early 2000s, when the GPS arrived. Everyone assumed that it would be a boon to cab drivers and they would be thrilled. What happened was more complicated.

For decades, London taxi drivers had earned their licences by passing an exam called The Knowledge. It was a two-to-four-year process of memorising 25,000 streets, 320 routes, and thousands of points of interest across the city. While the process was tough, earning the licence gave cab drivers a sense of identity and fulfilment. They could proudly claim: “I know this city like no one else does.”

When any driver with a smartphone could navigate the city, that identity of knowing the city better was gone. Cab drivers had to rebuild and, in some cases, rediscover the value they brought — value in the form of safety, judgement about picking routes, and a friendly demeanour. Things that a GPS cannot give you.

The Camera and the Portrait Painter

If you go a little further back, when photography arrived, you will see a similar pattern. Portrait painters were confronted with a machine that could capture a human likeness in minutes. No one would sit for days anymore to have their portrait made.

Painters and artists had to rebuild the value they brought to the table. They reinvented painting in such a profound way that we are still living it. Different genres like Impressionism, Abstraction, and Modern art emerged. The painters who thrived stopped competing with the camera and started asking what painting was actually for.

The Industrial Revolution and the Craftsman’s Pride

In the 1800s, the industrial revolution posed the same question to craftsmen. The weaver, the cobbler, the furniture maker — their mastery lived in their hands. The pride lived in the object created.

When factories could produce in a day what a craftsman produced in a week, the deeper loss was not income. It was the self-identity.

Same pattern in another century! In the fifteenth century, when the printing press arrived, it changed the cost of and access to books. It felt like an economic change, but it was deeper. It redefined how knowledge was transmitted and made accessible.

For scribes, it was not just the loss of jobs but a deeper question: if the machine can do the copying, what exactly am I here for?

AI Is Asking the Same Question — Again

Look closer and you realise AI is posing the same question to professionals everywhere today. Every role will have to go through some version of the history above.

Some roles will have to rediscover the value they bring — as cab drivers had to. Some roles will have to reinvent themselves — as the artists did. Some roles will need a transformation — as the scribes had to.

We tend to frame AI as a technology change. We would do well to remember it is also a question of personal identity and the value we bring. The conversation on AI is less about job security and more about these harder-to-navigate transitions.

Let’s begin asking the right questions over the next few posts to understand this identity aspect better. What am i most proud of in the work i do? What am i valued for in my teams? Where do i get my greatest joy from?

The answers will evolve. Getting the questions right is the task before us. The tool is changing, yes! The real shift, as it has always been, is about you.

The Software Stopped Waiting for You

Deepa Premkumar — Thu, 14 May 2026 04:31:03 GMT

Remember a time when we had big server rooms in offices? 25 years ago enterprise software lived in server rooms. Big physical machines humming in corporate basements and they looked like this.

Then, The Cloud Revolution Happened.

Companies like Salesforce, Workday, ServiceNow and hundreds of others made on-premise servers look slow, expensive and even absurd. Within a decade, the question shifted from ‘Should we move to cloud? To ‘Why haven’t you already?’.

Now the cloud + software that won the battle is facing its own version of the same revolution with Agentic AI. SAAS has started to look slow, expensive and even a bit absurd.

What actually happened?

At its developer conference in San Francisco earlier this year, Salesforce announced something they called Headless 360. The name is technical. The idea is simple: everything Salesforce can do — every workflow, every data record, every business rule — is now accessible via an API to AI Agents. No human needed to click through menus and to access data.

Salesforce co-founder Parker Harris put it as plainly: “Why should you (humans) ever log into Salesforce again?”

Here is an example, if a company had to dig the ground to lay a foundation for a construction -

25 years ago, Companies had to buy shovels and digging equipment to get work done (On-Premise servers).
Next came a time when companies could just rent the shovel and digging equipment (On – cloud). They did not have maintain the shovel, get it repaired etc. They just paid for usage. Of course, their employees still had to do the digging.

Now with Agentic AI, companies no longer need to rent the shovel and in many cases not have their employees dig too. They can just pay for the final outcome of having the ground dug up. Agentic AI is able to identify the best way to dig and get it done even before any employee has come to work.

The people who saw it coming

The Salesforce announcement is not new. Many people saw this change coming and even implemented it. Jason Lemkin, one of the most respected voices in the SaaS world, didn’t wait for a conference announcement to say this out loud. His own company runs an AI Agents for Marketing and Customer Service. They read and write the CRM continuously. His human employees barely log in. “The agentic future isn’t coming,” he wrote. “It’s here. And leaders who wait lose 12 to 18 months of compounding advantage.” The takeaway here is that every other SAAS vendor has a max of 12-18 months to respond. Headless model will become the new default.

Satya Nadella, CEO of Microsoft, probably predicted this much earlier in 2024 – where he said “business applications that are a bunch of databases with a bunch of business logic will collapse….The agent will orchestrate across multiple SaaS applications“. The SaaS model as we know it—where humans log into specific, siloed dashboards (like a CRM or HR tool) to manually click buttons—is becoming obsolete.

This change is reflected in the money trail as well!

We all have bought software based on number of users some point. This seat-based pricing — the model where companies pay per employee who uses a tool —according to Bain’s B2B report has dropped from 21% to 15% of the market in just twelve months. Gartner projects that by 2028, at least 40% of enterprise software spend will have shifted away from seat-based models entirely. IDC predicts that by 2028 – pure seat-based pricing will be obsolete.

In case you want to understand this better - Pranav Pathak at Builder Lab has a useful explainer on the revenue models for SAAS companies here.

The question nobody is asking out loud

Yes, Technology has moved ahead in leaps and bounds. Yes, the money is following technology as you can see from the revenue and projections. But what about the employees? How will they manage this transition where they no longer need to operate the software to create outcomes??

When the software stops needing you, what are you for?

This is where most technology transformation conversations go quiet. Not because the technology is too complex, but because the human question is too uncomfortable.

Our hypothesis is, most business leaders assume the announcement is the change. A platform goes live, a feature gets deployed, new technology is rolled out. However change and transition takes months/years and consistent effort.

William Bridges, one of the most important thinkers in organisational change, drew a distinction that most leaders never make. He separated change (the external event) from transition (the internal psychological journey) people go through in response to it.

Change is situational. Transition is psychological. You can flip a switch on a platform in a day. It takes far longer for people to let go of who they were before.

Most transformations fail not because the technology doesn’t work, but because leadership manages the change and ignores the transition. They deploy the software and declare victory. Meanwhile, employees are still running their own version of the pipeline report/other reports on the side, in good ol’ excels. Not because they don’t trust the agent or AI. Because they haven’t yet figured out who they are without making the report.

And this is where loss aversion enters the picture. Psychologists Daniel Kahneman and Amos Tversky found that people feel losses more intensely as gains.

Telling an employee that the agent will save them three hours a week is useful. But it won’t get them to whole heartedly adopt the new technology. That is because the unspoken loss, of the visible proof of their competence, of the thing that made them feel on top of things, that loss is more intense (than the gain of productivity).

Leaders who frame agentic AI purely as efficiency gain will keep being surprised by the friction they encounter. The loss needs to be acknowledged before the gain can be accepted and enjoyed.

This isn’t about technology. It is about identity. And it is a part of the transformation that often gets ignored.

Every big shift has done this

This has happened with every technology transformation. Humans have had to re-craft their identity time and again. When a core skill ,the "thing" you are known for gets automated, it does not just change your workflow. It changes your identity, the narrative you tell yourself.

Think of accountants when calculators and spreadsheets became common. The ‘thing’ they were good at (mental calculations) was automated. It did not just change their workflow. They had to create new identities for themselves. They stopped being the person who enters and calculates the data, and became the person who decides what the data means.

Think of travel agents in the early 2000s. Earlier they were information brokers. they had access to fares, routes, and availability that ordinary people did not. When Expedia and booking.com made that access available to all, their core value proposition evaporated overnight. The ones who thrived were those travel agents who rebuilt their identity as experience curators — specialists in complex itineraries, luxury travel, or niche destinations where human judgment, relationships, and taste still mattered. Identity moved from information access to curation and care.

The pattern is always the same. The technology arrives faster than the identity and narratives can adapt. The people who thrive are not those who resist, or those who blindly automate everything. They are the ones who move up the value chain .

So what do you actually do?

The question for every organisation right now is the same: are you actively helping your people make that move? Or are you simply deploying the technology and hoping they figure it out?

The organisations that will come out ahead are not the ones that move fastest. They are the ones that bring their people with them. Leaders who help employees recraft their identities by redefining what they value, is a good place to start.

Organisational Singularity: Explained Five Ways

Deepa Premkumar — Thu, 16 Apr 2026 06:00:56 GMT

In our last piece we introduced Organisational Singularity. The point at which an organisation’s AI shifts from tool to nervous system. Learning and improving continuously without waiting for human direction.

One responder asked, can you explain this to a 5 year old? Because that is a true proof of understanding.

We took the challenge further. So here it is. The same idea of Organisational Singularity, explained to 5 different groups - a five year old, a college student, a new hire, an L&D Head and a CEO.

To a 5 Year Old

You know how you get better at drawing the more you practice? You draw something, your teacher gives you a few more ideas and you try again. It comes out a little better each time.

Now imagine a whole school that did that on its own. Every classroom, every teacher, every lesson — all learning from each other at the same time. A student and teacher both learn through every practice in the school, not just from their own class.

That’s what happens when a company reaches Organisational Singularity. The whole organisation learns together — and it never stops practicing.

What you can try now to be ready for Singularity:

Keep practicing your drawing. Because the companies of the future will still need people who create things, imagine things, and care about things. The machines get better at remembering. You get better at wondering.

To a College Student

You have probably used AI to help with an essay or summarise a paper. That is using AI as a tool — useful, but waiting for you to direct it.

Now imagine a company where AI isn’t sitting in individual laptops waiting to be asked something. It is woven into everything — connecting what the sales team learns to what the product team builds. Connecting what customers say to what operations does next. Every decision feeds the system. Every outcome teaches it something. Nobody had to schedule that learning. It just happens.

That is the difference between a company using AI and a company that has hit Organisational Singularity. One is using a very fast calculator. The other has built something that thinks alongside it.

What you can try now to be ready for Singularity:

The best thing you can do right now is run an experiment. Pick one project — academic, personal, creative, anything — and try to build a learning loop around it. Document what you learn. Feed the outcomes back in. See if the project starts improving faster than you expected. Use AI not just as a tool to complete tasks but as a system that gets better the more you feed it.

You won’t be building Organisational Singularity. But you’ll understand the feedback loop from the inside. Which is worth more than any certification when you walk into a job interview and someone asks how you work with AI.

To a New Hire

You’re in your first few weeks. You’re trying to figure out how the company works! You are trying to find where the documents and information is saved. How decisions are made, and what are some unwritten rules of the company.

Typically, organisational or Institutional memory is rarely documented or easy to find for someone new. Many a times it simply lives in the heads of the more experienced employees.

In an organisation approaching Singularity, something structurally different is happening. The institutional memory is no longer held only by people — it is alive and visible in the system itself. In these companies, AI is quietly connecting decisions to outcomes. AI is surfacing patterns no individual would notice. And feeding that learning back into how the organisation operates next time.

What you can try now to be ready for Singularity:

Pay attention to how this organisation learns. That will tell you more about its future than any strategy deck. The skills that will matter most here aren’t just execution skills. They are the ability to ask the right questions of a system that can now answer faster than any manager could.

To an L&D Head

Think about how a new employee learns today. They go through onboarding, attend programmes, shadow colleagues, make mistakes. They get feedback, and gradually build up knowledge over months and years. That learning is real — but it is individual. And it doesn’t automatically transfer to the next person who joins.

Now imagine if every lesson that employee learned — every mistake, every good decision, every customer interaction — was instantly captured. It was connected to outcomes, and fed back into how the whole organisation operates next time. Not stored in a document somewhere. Actually applied. Automatically. By the system itself.

That is Organisational Singularity. The organisation stops learning the way individuals learn. Slowly, manually, one person at a time — and starts learning the way a well-designed system learns. Continuously, collectively, and without waiting for anyone to schedule it.

What you can try now to be ready for Singularity:

This raises an uncomfortable question for the L&D function. If the organisation is already learning continuously through its systems, what exactly is your job?

The L&D function that survives Organisational Singularity is not the one with the best LMS. It is the one that understood that its real job was preparing humans for a new kind of collaboration. Every person in your organisation needs to understand — at the level appropriate to their role — how to work alongside the intelligence layer productively. This is the new organisational fluency.

To a CEO or Senior Leader

Think about why decisions land on your desk.

Rarely because someone below you lacks authority. More often because they lack the information, the context, or the judgment to make the call confidently. A regional manager escalates a pricing decision because they don’t have visibility across markets. A new hire doesn’t even attempt the decision. they have not spent enough time to know what good looks like yet.

So decisions travel upward. Your calendar as CEO/CxO is full not because you are the only one who can decide. But because you are often the only one who has enough to decide well.

Now imagine an organisation where that information problem no longer exists at any level. Where a regional manager about to make a pricing call is shown — in real time — what similar decisions produced in comparable markets. Where a new hire, is guided by a system that surfaces the institutional wisdom they haven’t had time to accumulate yet.

That is Organisational Singularity. The system doesn’t make the decision. But it arms every level of the organisation with the kind of informed context that previously only lived at the top. Decisions stop travelling upward out of necessity. They get made where they happen — faster, better, and with less dependence on any single person’s judgment.

What you can try now to be ready for Singularity:

This changes your job more than anyone else’s in the organisation. You are no longer the answer to an information problem. The system handles that. What you become is the answer to a values problem. The intelligence layer can recommend, optimise, and surface. It cannot decide what the organisation stands for. It cannot make the calls that require moral judgment, long-term thinking, or the kind of wisdom that no feedback loop can replicate. That remains irreducibly yours.

The question isn’t whether Organisational Singularity is coming. Are you building toward Organisational Singularity — or managing the bottlenecks until you can’t.

In closing

The same concept. Five different moments of recognition.

What’s interesting is that Organisational Singularity doesn’t feel the same from every seat. It feels like wonder to a five year old. Like a project to a college student. Like a mentor to a new hire. An existential question to an L&D head. And maybe a full yet different calendar to a CEO/CxO.

But underneath all five — the same shift is happening. The organisation stops waiting for humans to move knowledge around. It starts doing it by itself.

Organisational Singularity: What It Is and Why It Matters

Deepa Premkumar — Sun, 05 Apr 2026 07:01:40 GMT

Ray Kurzweil, the futurist who predicted the rise of the internet, the smartphone, and AI’s dominance decades before they arrived, has long spoken of a Singularity — a point in the future where AI would equal and then surpass human intelligence. Not through a dramatic event, but through a self improving loop of AI. He predicted this would happen by 2045.

Today we are seeing clear signs of the feedback loops which are helping AI get better faster. Remember the first version of GPT we used in 2023-24, it was exciting but was not very useful. It seems like a pale comparison of the AI models and agents we use today. AI has moved away from needing constant prompting to be able to run commands and make decisions independently.

As individuals and teams get better with AI - what about organisations? What happens to organisations that manage to adopt AI differently? How much smarter do these organisations become? What if there was a self improvement loop for organisations too, driven by AI.

What if there was a possibility of an Organisational Singularity?

What Is Organisational Singularity?

Organisational Singularity is the point where data —>decisions —> and learning connect in a continuous loop that improves itself over time, with or without anyone consciously directing it. It arrives for organisations when AI stops being something employees use as merely a tool and starts being something the organisation thinks with. Not a system that replaces human judgment. A system that makes the organisation’s collective judgment sharper, faster, and self-reinforcing.

This is different from what Kurzweil described. He was talking about machines surpassing individual human intelligence. Organisational Singularity is — the moment an organisation stops depending only on humans to connect the dots, transfer the knowledge and starts doing it by itself.

Here is a comparison and an example:

Any organisation today is smarter than any ‘single’ individual within them. A hospital knows more than any one doctor. A bank knows more than any one analyst. But organisations today are dependent on people showing up, communicating effectively and transferring knowledge. It is fragile and slower. It learns through meetings, reports and decisions made by humans.

Consider a hospital in the future, where patient intake data feeds a diagnostic AI. This surfaces patterns that improve treatment protocols, which generate better outcome data. Which then trains the diagnostic model further. Individual clinicians and doctors are not replaced — in fact their expertise becomes sharper because the system surfaces what experience alone would take years to accumulate. But crucially, the organisation is now learning between decisions, not just because of them. The institution itself is getting smarter — continuously, without a meeting being called or a strategy being announced.

This is Organisational Singularity in action. And what makes it singular is this: the loop doesn’t stop. It compounds.

Why is organisational singularity becoming possible now?

The main reason why organisational singularity is becoming possible - is the shift from intelligence scarcity to abundance.

For over a century, organisations were designed around the constraint that intelligence was scarce and information was expensive. Hierarchy in companies solved this by centralizing knowledge on a need to know basis. Ronald Coase won a Nobel prize for articulating this in 1937 in his paper ‘The Nature Of The Firm’.

Today digital infrastructure and AI have collapsed external transaction costs and the marginal cost of intelligence. In many cases, it is now getting more expensive and slower to perform a task through a corporate hierarchy than to use AI or external digital networks.

The organisational singularity occurs when firms redesign themselves around intelligence abundance. Humans are not replaced in these firms but they lower internal friction and focus on possibilities and speed rather than merely control.

Principles of Organisational Singularity

Three principles help in understanding Organisational Singularity better.

The Compounding Principle

Humans have always been good at visualising and seeing improvements right in front of them. Think of a weight loss or fitness program you have been on. The first two weeks of exercise feels brutal with no visible return. Maybe by month two you start noticing change and it is month six when people start noticing! By then most of us have given up, because we cannot visualise the changes. Nothing dramatic happened in any single week. The body just kept compounding quietly - and the results became visible much later.

AI Singularity is following this exact pattern. Generative AI in 2018 looked interesting but modest. By 2023 it had made a splash and acquired millions of users. By 2025 it was rewriting entire industries. Nobody saw the magnitude coming — not because the signals weren’t there, but because our brains discount exponential curves.

Organisational Singularity works the same way. The organisations quietly building self-improving feedback loops today don’t look dramatically different from the outside yet. Their annual reports look similar. Their headcounts are comparable. But underneath, something is compounding — better data feeding sharper AI, sharper AI producing better decisions, better decisions generating richer data. Round and round, tightening with each cycle.

The compounding happens below the surface before it shows up in results. By the time it’s visible, the gap is already very difficult to close.

The Systems Principle

“The shift is not from people to AI. It is from AI as a tool to AI as connective tissue”.

Most organisations today use AI the way earlier generations used early computers — as a faster way to do existing things. Drafting faster. Analysing faster. Searching faster. The underlying operating model remains the same. Humans still make the calls. Processes still run in silos. Institutional knowledge still lives in people’s heads and disconnected systems.

Organisational Singularity describes a different configuration entirely. One where AI becomes the connective tissue of the organisation — linking departments, decisions, data, and learning in real time. Where insight generated in one corner of the business automatically informs another. Where the organisation doesn’t just store knowledge, it applies it continuously.

This is the operating system shift — and it’s why two organisations with the same AI tools can be on completely different trajectories.

Evolution (Not Extinction Principle)

“The question is not whether people will be replaced. It’s what people will become.”

The most persistent anxiety around AI — at the individual level and now at the organisational level — is replacement. If the system learns and improves by itself, where does that leave the people inside it?

Ray Kurzweil’s answer at the individual level was always the same: humans don’t disappear, they evolve. The relationship between human intelligence and machine intelligence becomes collaborative and over time deeply intertwined.

The same logic applies inside organisations. Look at history to see the pattern below:

Before electricity, factory managers managed physical muscle — the effort and output of human bodies. After electrification, they managed machines and processes. The role didn’t disappear. The nature of it transformed entirely.

At Organisational Singularity, the transformation is similar. Leaders and managers don’t get displaced by the learning system — they move up a level. They shift from managing processes to managing the intelligence layer itself. Setting values and direction. Making judgment calls that require context, ethics, and humanity that AI cannot supply.

“Leaders in the AI era will shift from managing processes to managing and the intelligence layer itself and orchestrating the changes”

Organisational Singularity is Underway Already

Organisational Singularity is not inevitable. Not every organisation will cross this threshold, and crossing it is neither automatic nor guaranteed to be easy.

But for the organisations quietly building these systems right now — connecting their data, closing their feedback loops, rewiring their operating models — something is already underway. The compounding has begun. And like all exponential processes, it will look modest right up until the moment it looks remarkable.

Organisational Singularity has possible already begun its improvement loops. Through small experiments, through learning, through lobsters*!

*The lobsters are real. And they are already learning (Lobster)

From Execution to Insight: The Shift AI Is Making Possible

Deepa Premkumar — Wed, 11 Mar 2026 08:12:07 GMT

Five hundred and forty-two. That is the number of leaders currently on development journeys through Flyntrok’s platform — a proprietary tool that weaves organisational context with nudge theory to keep leaders engaged in their growth over time. Across those plans, one pattern is nearly universal. Strategic and long-term thinking shows up, again and again, as the competency that needs the most development focus.

This is not just what we see internally. DDI’s Global Leadership Forecast 2025, spanning over 50 countries and 24 industry sectors, finds that leaders themselves name setting strategy as their greatest development need — with only 22% of HR teams currently prioritising it. The Centre for Creative Leadership reports the same finding. The gap is persistent, and it is widespread.

Here is what struck us, though. The leaders on our platform are capable and successful. This is not a capacity problem. So, we began to ask a different question: what if strategic and long-term thinking is not underdeveloped because leaders cannot do it — but because the conditions around them have made it easy not to? And what if AI is about to change that?

Two Ends of a Spectrum

Picture a spectrum.

At one end is execution: tasks, outputs, deliverables — things you can point to at the end of the day and say, I did that.

At the other end is insight: judgment, sense-making, sitting with a problem that does not have a clean answer, working with the complexity of people and context.

Most leaders, when asked honestly, know which end of that spectrum absorbs most of their time. And most know their role asks for more of the insight end than they typically give it.

The powerful question is not whether leaders know this. It is why execution pulls so naturally ahead.

Why Execution Pulls Ahead

Execution has a lot going for it. It is urgent, visible, and satisfying. It produces something tangible. And in most organisations, the systems — the metrics, the meeting rhythms, the performance conversations — are built around outputs. As Cal Newport has observed, organisations tend to measure and reward what is easy to see: tasks completed, emails answered, decisions made. One thoughtful strategy conversation is harder to point to than ten smaller problems resolved.

Insight work asks for something different. It involves sitting with questions that do not resolve quickly. It requires comfort with uncertainty, and it can feel exposed in a way that execution rarely does. Not because leaders lack the capability — but because the conditions for that kind of thinking have not always been built into the working day.

The systems were not designed to be hostile to strategic thinking. They were designed around what was easiest to see and measure. And that quietly shaped what got prioritised.

What AI Is Changing — And What It Is Really Asking

This is where generative AI becomes genuinely interesting. Not just as a productivity tool, but as something that surfaces a deeper question about role.

The choices made at that point will define the role. Whether the recovered time gets filled with more execution, or invested in judgment calls, sense-making, and harder conversations — that is an identity question as much as a time management one.

The shift AI is prompting is not just a behaviour change. It is an invitation to recraft the role — to move from being the person who gets things done to being the person who thinks about what is worth doing and why.

Recrafting the Role

What lives at the insight end of the spectrum is not harder work. It is more human work.

MIT Sloan researchers recently mapped the capabilities AI is least able to replicate, naming them the EPOCH framework: Empathy, Presence, Opinion and Judgment, Creativity, and Hope. Significantly, they refuse to call these soft skills. A hard skill like solving a maths problem, they point out, is comparatively easy to teach. EPOCH capabilities are far harder to develop — and far more distinctly human.

These are precisely the capabilities that define the insight end of the spectrum. The ability to read what is really going on in a room. To make a call when no option is clean. To sit with a team navigating change and offer something more useful than a process. None of this is new to leadership — but for many leaders, it has been the part of the role that got squeezed by everything else that needed doing.

AI is not asking leaders to become different people. It is creating the conditions for them to become more fully the leaders they were already meant to be.

Recrafting the Organisation

This question does not sit with individuals alone. It is also an organisational design question.

If generative AI is compressing execution work, but the systems still measure and reward execution-style outputs, the shift will not happen at scale — regardless of individual intention. Organisations need to ask honestly: is there genuine space for ambiguity and strategic conversation? When insight-level thinking does happen, is it seen, valued, and reflected in how performance is understood? Redesigning what gets measured is as important as redesigning what gets done.

An Invitation

The capacity for judgment, sense-making, and strategic thinking has been there all along. What generative AI is changing are the conditions. With execution taking less time, the question of what to do with the space that opens becomes both more urgent and more possible to answer well.

That feels, to us, like an opportunity worth taking seriously.

This is the third post in a five-part series on how generative AI is redefining professional roles — not just making us faster. Read the opening post here: When the Tool Changes, But the Real Shift Is About You.

References

DDI — Global Leadership Forecast 2025 (Link)
Center for Creative Leadership — The Leadership Gap (White Paper)- (Link)
MIT Sloan — EPOCH Framework (Loaiza & Rigobon, 2025) (Link)
Newport, Cal — Deep Work: Rules for Focused Success in a Distracted World (Grand Central Publishing, 2016)

When the Task Disappears, Who Are You?

Deepa Premkumar — Sun, 08 Mar 2026 10:40:55 GMT

Picture a software engineer who has spent fifteen years defining herself through her code. The elegance of her logic, the efficiency of her solutions, the quiet pride of a clean build. Then, gradually, and then suddenly, AI begins writing the code. Not all of it. But enough of it that she finds herself staring at a screen one afternoon, wondering something she has never wondered before. If I am not the one writing this code, what exactly am I here for?

That question — uncomfortable, disorienting, and surprisingly common right now — is not a productivity problem. It is an identity problem. And the first thing worth saying about it is this. Identity questions are not new. It is not a sign that something has gone wrong with you and us. It is a sign that something very human is being surfaced by a very powerful technology.

A Century of Thinkers Who Saw This Coming

For over a century, some of the sharpest minds in psychology and sociology have been circling the same insight. Identity is not a fixed thing you are born with and carry through life unchanged. It is something you construct, perform, and rebuild — largely through what you do, who you do it with, and how the world responds to you.

William James said it first, in 1890, when he argued that we have multiple selves depending on our context. George Herbert Mead built on it, showing that the self is an ongoing internal dialogue between our impulses and the social world around us. Erving Goffman made it vivid. He described professional life as a performance. With a front stage identity we present to the world and a back stage self that exists underneath it.

Erik Erikson gave us the language for the moment the performance is disrupted: “identity crisis”. Not a pathological problem but a necessary passage that opens when the old story no longer holds.

What all of them were pointing to, in their different ways, is the same thing: we construct our sense of professional self through our tasks. Which means when the tasks change, the identity question opens up whether we are ready for it or not.

The Confusion Between Means and Ends

Here is the deeper problem. Most of us, without ever consciously deciding to, have confused the means for the end. The software engineer was never really the person who writes code. She was always the person who solves complex problems. Translates human needs into functional systems, and makes something work that did not work before. The code was infrastructure. It was the means. The judgment, the problem framing, the ability to ask the right question of a system — that was always the point. But because the task and the role were never separated, the task became the identity. And now that AI is absorbing the task, it feels like the identity is being absorbed too.

It is not.

But our identity does need to be redrawn and recrafted in the AI era.

You Cannot Think Your Way Into a New Identity — You Have to Act

This is where Herminia Ibarra’s work becomes essential. Ibarra, one of the most important thinkers on professional identity and career transitions, argues that you cannot think your way into a new professional identity. Introspection alone will not get you there.

You have to act your way into it. Try things, take on new kinds of work, experiment with a different version of yourself in practice. And let the new identity form through doing rather than through reflection alone. The engineer who spends six months thinking about what her role means in an AI world will arrive at the same uncertainty s/he started with. The engineer who spends six months using AI to tackle bigger, more strategic problems — and pays attention to what s/he learns about herself in the process — will find the new identity forming almost without her noticing.

This is not always easy, alebit very important. Redrawing the arc of your professional identity is not a weekend exercise or a line in a performance review. It deserves honesty, deliberate effort, and in many cases, support.

The disorientation is real. The grief for the old version of the role is real. None of that should be minimised.

This Is Not Just a Personal Question — It Is An Organisational One

In our work, we see professionals across industries and geographies, struggling with this. Sometimes it is a question being asked across entire teams and functions. And most organisations are not designed for it. Job descriptions are still written around task clusters that AI is dissolving. Performance frameworks still reward the execution of things that will soon be automated. Hiring decisions are still anchored in technical skills whose half-life is shrinking by the month.

The organisations that will navigate this well are not necessarily the ones moving fastest on AI adoption. They are the ones creating deliberate space for their people to ask the identity question. To examine what their roles are actually for, to separate the task from the purpose it was always meant to serve. And to redesign around that purpose rather than around the task.

Slowing Down To Meet the Change

The technology is moving faster than the conversation about what it means. This post is an attempt to slow that down for a moment — not to resist the change, but to meet it with the clarity it deserves.

When the Tool Changes, But the Real Shift is About You

Deepa Premkumar — Sun, 08 Mar 2026 09:09:37 GMT

Every Generation Has Been Here Before — But This Time Feels Different

Every generation of professionals has lived through a technology that changed what they do. The spreadsheet and computer replaced the paper ledgers. Email replaced the memo. Each technology wave automated a task, sped up a process, removed a friction. And each time, we adapted — updated our habits, learned a new tool, and largely carried on defining ourselves the way we always had.

Generative AI feels different. Not only in scale or speed, but in the specific kind of disruption it creates. For the first time, the technology is not only changing what we do. It is asking us who we are at work. That is a different conversation entirely. And it is one worth having deliberately rather than stumbling into.

Three Reasons This Moment Is Unlike Any Before

So why is this moment different? Why is generative AI creating the conditions for role redefinition in a way that previous technologies did not?

It speaks your language.

Every tool before this required you to learn its logic. Formulas, code, software syntax. You had to translate your thinking into the tool’s terms. Generative AI inverts that relationship — you bring your natural language, your context, your intent, and it meets you there. When you do not have to learn any specific skill to use this technology, something interesting opens up. You suddenly have space to ask: what am I actually bringing to the table?

The speed makes your old story collapse.

When something that once took you two days of work now takes four minutes, the cognitive dissonance is too large to ignore. You cannot simply update your role definition at the margins. The gap is too wide. Previous technologies made things incrementally faster, or altered some aspects of the role. Generative AI makes speed hardly a criteria. And it is that categorical shift that forces the identity question, whether you are ready for it or not.

It removes the skill barrier as the last excuse.

With earlier technology, not knowing how to do something was a legitimate professional boundary. “I am not a data analyst” was a reasonable thing to say. Generative AI quietly dismantles that defense. When the technical barrier collapses, the question that remains is both clarifying and uncomfortable: what is it that only I can bring? That question — sitting there with no easy deflection — is what makes this moment different.

The People Who Are Moving Forward

Across our work, and in conversations with leaders across industries, we keep seeing the same pattern. The professionals who are most energised right now are not necessarily the ones who have mastered the most tools. They are the ones who have accepted the invitation — and started exploring what it means to respond to it.

They are sitting with questions like: what is the difference between the tasks I do and the role I hold? Where on the spectrum between execution and insight am I spending my best thinking? Am I using AI to do things better, or to question whether I am doing the right things at all? And what becomes possible when the barriers I once bumped against simply are not there anymore?

None of them have fully cracked it. Neither have I. But the willingness to ask the question at all seems to be what separates the people who are moving forward from the ones who are waiting for the dust to settle.

This Is Bigger Than a Tool Upgrade

This is not a small shift. It touches how we see ourselves, how we hire, how we build teams, and how we invest in our own growth. And it is happening faster than most organisations are prepared for.

Over the next few posts, I want to go deeper into four lenses that help make sense of this shift — the difference between tasks and roles, moving from execution to insight, questioning what we do rather than just doing it better, and discovering what becomes possible when old barriers fall away.

But the place to start is here: generative AI is not upgrading your toolkit. It is upgrading the question of what your role is for. The professionals and leaders who recognise that early will not just be faster. They will be further ahead.

The 11 Pitfalls That Keep Organisations Stuck in AI Enhancement Mode

Deepa Premkumar — Wed, 11 Feb 2026 08:33:14 GMT

Most organisations are using AI. Very few are using it well. Many organisations deploying AI are stuck in what we call “enhancement mode” — using a general-purpose technology to do old things slightly faster. They are automating bad processes instead of reimagining good ones.

And they are leaving transformative value on the table.

To be fair change and transition is not a linear, simple path. It is a journey of experiments, detours and crossroads. Managing this AI transition and change will be more about the organisation’s ability to think differently rather than merely spend. Because the difference between organisations that extract real value from AI and those that do not is not the technology they buy. It is the organisational decisions they make — or fail to make — at four critical stages.

The Four Stages of AI Adoption Within Organisations

AI adoption is not a one-time rollout. It’s a cycle that organisations move through repeatedly — every time they tackle a new use case, every time capabilities advance. Think of it as a loop, not a line. And at each turn of this loop, there are four stages where decisions determine whether you stay stuck or break free.

Stage 1: Procurement

Before you deploy a single tool, you make foundational decisions that set everything in motion. How do you position AI internally — as a threat or a strategic capability? Do you start with real problems or chase shiny demos? Do you build internal muscle or outsource your transformation to consultants? Get procurement wrong, and every stage that follows inherits that original mistake.

Stage 2: Deployment

Two decisions determine whether AI spreads virally through your organisation or dies quietly in the pilot phase. Do you treat AI like another software rollout — install it, train people once, measure logins? Or do you recognize that AI requires fundamentally different deployment thinking? And who owns the AI agenda — IT alone, or the people who actually do the work?

Stage 3: Evaluation

You cannot improve what you do not measure. But measuring the wrong things keeps you stuck. Are you tracking adoption metrics that look impressive in a dashboard but tell you nothing about value? And when early results underwhelm — as they inevitably will with any general-purpose technology — do you pull the plug too quickly, or do you have the patience to let transformation unfold?

Stage 4: Growing

AI capabilities do not stand still, and neither can your organisation. Have you redesigned workflows and structures, or just layered AI on top of what already existed? Are you building capacity for tomorrow’s AI, or optimizing only for today’s? And critically, are you thinking about the humans in this equation — the apprenticeship pipelines that are breaking down, the career arcs that need redrawing?

The 11 Pitfalls of AI Adoption Within Organisations

Across these four stages, we have identified 11 specific pitfalls — predictable traps that organisations fall into, often with the best of intentions.

Here is the map:

Pitfall 1: Positioning AI as a Threat
Pitfall 2: Buying Tools Before Defining the Problem
Pitfall 3: Outsourcing Transformation Instead of Building Internal Muscle
Pitfall 4: Treating AI Like a Plug-and-Play Software Rollout
Pitfall 5: Banning, Centralising, or Outsourcing AI to IT
Pitfall 6: Sticking with Existing Metrics
Pitfall 7: Giving Up Too Quickly
Pitfall 8: Not Redesigning Workflows and Structures
Pitfall 9: Not Building Capacity for Tomorrow
Pitfall 10: Broken Apprenticeship Systems and Talent Pipelines
Pitfall 11: Not Helping People Redraw Their Arc

Why This Matters Now

There is a pattern worth noting: these pitfalls compound. Position AI as a threat (Pitfall 1), and you guarantee resistance during deployment (Pitfalls 4 and 5). Measure the wrong things (Pitfall 6), and you will almost certainly give up too quickly (Pitfall 7). Skip workflow redesign (Pitfall 8), and no amount of investment in talent pipelines (Pitfall 10) will save you.

What’s Next:

Over the next several posts, we will unpack each of these 11 pitfalls in detail. Each post will include the core insight, a real-world narrative of what goes wrong, a counter-example of what going right looks like, and practical guidance for breaking free.

We start first with the pitfall that poisons everything downstream: what happens when your organisation — deliberately or by default — positions AI as a threat to the people who are supposed to use it.

Images are AI generated: Hallucinations and typos to be excused!

This is the first in a series of 12 posts exploring the organisational pitfalls of AI adoption. Next up: Pitfall #1 — Positioning AI as a Threat.

The AI Reading Paradox

Deepa Premkumar — Fri, 14 Nov 2025 06:23:43 GMT

The habit of reading has been declining over decades. Reading struggles to compete with the videos and feeds! Will Generative AI be the death knell of reading? This is what we explore in this edition.

No doubt, Artificial intelligence is transforming how we read. But the impact isn’t straightforward. The same technology that liberates one reader is imprisoning another. I am sharing two real lived experiences with Generative AI and reading. The difference isn’t in the tool—it’s in how we use it.

The Liberator: AI as Access

A friend of mine, Maria (name changed), a UCL graduate student in the UK, faced an uphill task. reading a single academic paper took her four hours. Not because she wasn’t intelligent, but because English wasn’t her first language. Academic writing, jargon, and complex sentences made it so much harder.

She started using AI and everything changed. The tool didn’t read for her—it helped her understand complex passages. AI explained terminology, and adjusted text complexity when she hit walls. Within months, her comprehension soared.

Research backs this up. A 2025 study in Smart Learning Environments found that AI intervention led to “significant improvements in reading comprehension and self-regulated learning behaviors”. With “increased attentiveness, participation, and motivation” among second-language learners.

For students like Maria, AI democratizes access—it levels a playing field that was never fair to begin with.

The Trap: When Convenience Becomes A Handicap

But there’s another story. My good friend’s son Kobe (name changed again) discovered a shortcut. Copy paste assigned readings into Chat GPT, get instant summaries. It helped him score well on quizzes and objective type questions. If a written assignment had to be submitted, he asked AI to write like an 8th grade student and submitted it. The system rewarded him—until it didn’t.

For the half yearly exams, the teacher assigned an original analysis in the classroom. Kobe struggled and couldn’t do it. He had never actually engaged with the ideas, never wrestled with arguments, never experienced that breakthrough moment of genuine understanding. He’d outsourced the struggle. And the struggle was where learning happened.

The data is stark. Research from Duke’s Innovation Co-Lab (2023) found that complete reliance on AI for reading tasks led to a 25.1% reduction in accuracy. Students who let AI do their thinking performed worse than those who engaged with study material. The convenience comes at a cost.

The Real Danger: Both Extremes

The more we looked at the various research on AI and reading.

One thing becomes clear - (both) refusing to use AI and over-relying on it can harm us.

Students who avoid AI tools entirely may struggle unnecessarily with inaccessible/tough material. Wasting brain power on language barriers rather than ideas.

But those who outsource all mental effort to AI never develop deep reading skills. Deep reading is what transforms information into understanding.

Striking the Balance: Three Practical Principles

Here are three principles for you to consider while reading varied content. These principles lend themselves to reading for work, leisure, knowledge or anything at all…

1. Engage First (The Diagnostic Principle)

Always attempt the reading yourself before turning to AI. Five to ten minutes of genuine effort will tell you whether you’re facing a language barrier or a conceptual challenge worth preserving.

2. The 50% Rule (The Effort Principle)

If AI could handle 100% of a reading task, intentionally do at least ``50% yourself (thumb rule/indicative). Use AI to check your understanding, not replace it—summarize first, then compare.

3. Remove Barriers, Not Thinking (The Purpose Principle)

Let AI handle surface-level obstacles like vocabulary, syntax, or translation. But keep the conceptual wrestling for yourself—that’s where understanding actually happens.

AI isn’t destroying reading—but it is revealing what reading is really for. It’s not about moving your eyes across words. It’s about the irreplaceable human act of thinking deeply about ideas.

Additional Research Links:

The Impact of AI on Students’ Reading, Critical Thinking, and Problem-Solving Skills.. Link

Reinforcing L2 reading comprehension through artificial intelligence intervention: refining engagement to foster self-regulated learning.. Link

Assumptions Shape Organisational Culture

Deepa Premkumar — Thu, 04 May 2023 10:20:47 GMT

"This team member sent me an email on a non-working day. It shows her commitment" said a business leader in conversation. This statement is not only assessing the team member. It says a lot about the beliefs and assumptions of this business leader too. Assumption and beliefs held closely by the team, shapes organisational and team cultures. In this edition of The OWL on Agile & Experiments, let us explore how assumptions shape organisational culture. We also discuss, how to spot and change assumptions, to shape your organisation’s culture.

Organisational Culture & Assumptions

Edgar Schein, defines organizational culture as "a pattern of shared basic assumptions that a group working together for a common goal has created. in learning to cope with the problems of external adaptation and internal integration." There are many many layers to this definition. For now, let us explore culture from the lens of shared assumptions made by the group.

Assumptions as we know from earlier blogs is our way of making sense of how things work. Assumptions help us make decisions and get things done. The other aspect worth keeping in mind, is that assumptions create a cycle. One that can be virtuous or vicious. It also is a dynamic, ever evolving cycle. How?

Assumptions influence how we think and therefore how we act. How we act further strengthens the wiring in our brains about how things work! Thinking and actions most repeated by the team members becomes the norm/ culture of the team.

Examples of assumptions impacting organisational culture

Assumptions by themselves do not damage organisational culture, but they shape it. The danger for organisations is when leaders and teams do not pause and reflect upon:

Whether the assumptions still hold good? Or has the need and context of the organisation shifted?
Whether the assumptions held create intended outcomes for the teams?

We see this at play in many organisations in many ways. For instance, a colleague joined a hierarchical organisation as a business head. She held different beliefs about hierarchies and access to leaders, due to her varied experiences. She set up an 'open-door' process, among other measures. Whereby whenever the leader's door was open, anyone in the team could walk up to discuss their ideas and issues. Many weeks later, she said the team was yet to walk through the door. The team members had not stopped to question if their assumptions about hierarchy needed to shift!

"Assumption is the enemy of communication" - Winston Churchill.

Zappos call centre employees share a strong belief that providing outstanding service will result in loyal customers. These beliefs and assumptions show up as observable behaviour. So much so that employees send potential customers to other retailers if Zappos doesn’t have the item in stock

Another much bemoaned change effort is that of diversity and inclusivity within organisations. Leaders set up diversity (gender/ ability/ nationality/ etc) hiring targets for teams to meet. Diverse talent does not always thrive within these teams. Assumptions and beliefs about the capabilities of the diverse talent recruited (most often held subconsciously and never discussed) gets in the way. These assumptions lead to minor and major acts of exclusion towards diverse talent. Thereby stifling their potential.

Shaping organisational culture through assumptions

Counter-intuitive as it may sound assumptions can indeed help business leaders shape and fashion the culture they wish to create in their teams. The secret sauce is to be able to spot assumptions which lurk right under the surface.

The ways to spot assumptions are many. Language, ladder of assumptions, 5 tools to spot assumptions, Going beyond the yes and more. In this edition we offer a few more tools and clues to spot shared assumptions held by teams.

This first one is possibly the simplest to spot and most effective when tried. Whenever you hear the words, 'this is how things are done here' - pause and probe further. Ask inviting, open ended questions to find out what shared beliefs and assumptions are making teams believe this statement.
Watch closely to what people pay attention to. Visible artefacts many a times exist because they represent shared assumptions and beliefs. Artefacts in turn shape assumptions and thereby culture.
Listen to the stories and anecdotes that are retold frequently. Stories are a powerful gateway to understand the culture of a team. They show up the inherent beliefs and shared assumptions the team holds

Spotting assumptions and changing them to shape the culture of your team is an ongoing endeavor. As Mr. Vivek Patwardhan says, "every team member impacts the culture of your team. It is a dynamic and evolving phenomenon. Think of it like a living system". Couldn’t agree more.

Agile Stories

This story is from Bank of America. The earliest part of this company dates back 240 years. They are one of the Big4 banks in United States roughly serving 10.4% of the country’s population. When Kenneth Lewis became the bank’s CEO in 1999, he realised the bank needed fresh approaches in services delivery. He also noticed something strange and decided to rectify it.

Have you noticed that most R&D departments exist in manufacturing companies? When Bank of America wanted to create new service concepts for retail banking, they realised they needed an R&D mindset. To make experimentation part of their culture and DNA, they took a few crucial steps. Listing a few (non-exhaustive list):

Set up an Innovation & Development (I&D) team.
Turned a set of its branches into labs, where they could run experiments with actual customers.
Created an idea portfolio and categorized them based on importance.

They have run several experiments till now. Refining the ways they learn and measure from these experiments too.

One experiment that stays with me, when the bank wanted to improve the wait time experience of its customers. Their hypothesis was improving the wait time experience of customers would lead to better revenue metrics in the branch.

One of the ideas to improve wait time experience of customers was to install TV screens in the branches, with details of BOA products being displayed. The team installed television sets across 20+ of its branches to test the idea. They measured both customer responses and the revenue figures over a few months.

The experiment showed that customer feedback had gone up however the revenue metrics had not. The team concluded that the returns for installing television sets across its thousands of branches would not be profitable. They decided to test other ideas for improving customer experience during wait time.

Being careful in how you define success of the experiments is crucial to ensure the right learning and outcomes.

'Build-Measure-Learn' To Grow

Deepa Premkumar — Tue, 28 Mar 2023 14:42:37 GMT

If I had to start a shoe store, even an online shoe store, I would have at least a few pairs of shoes in my stock. Wouldn’t you? The world's largest online store for shoes started without a single stock or inventory of shoes! Yes, Zappos used the principle of Build-Measure-learn (a principle of lean methodology) to build their online business from scratch. In this edition of the Agile OWL, let’s explore how we can reduce uncertainty and improve outcomes by using the build-measure-learn loop with our teams and in our lives.

Reducing uncertainty

Uncertainty and doubt are omnipresent or rather should be claim scientists and philosophers alike.

“The very foundation of science is to keep the door open to doubt.”

said Carlo Rovelli, a scientist.

“Doubt is an uncomfortable condition, but certainty is a ridiculous one”.
said Voltaire the 18th century philosopher.

The one that stays with me is a quote by Tony Schwartz the CEO of the Energy Project. Where he urges us to ‘Let go of Certainty’

He goes on to say "The opposite of certainty is not uncertainty. It is openness, curiosity, and a willingness to embrace paradox, rather than choose ... sides. The ultimate challenge is to accept things and ourselves exactly as we are, but never stop trying to learn and grow.”

In a complex and dynamic world that we live in currently, wishing away uncertainty is pointless. Rather looking to reduce uncertainty as we work is a better goal to have. Agile and Lean methodologies help leaders and teams achieve exactly that through experiments and the Build-measure-learn cycle.

Build-Measure-Learn Cycle

A ex-colleague of mine had a product idea which would solve a specific problem of internships in colleges. The problem? Students struggle to get internships and companies struggle to hire the right interns. Very excited to have found a true business need, he quit his job to devote more time to this business idea. In 12 months, he had designed the entire solution, written the code and had a final product ready!

He showed demo’s of the solution to colleges, but they were dis-interested. He decided to try ground up and invite students and companies to sign up directly, again finding lesser traction. While the feedback he was getting at this point was useful in modifying the product further. There was a catch. Unfortunately, he was out of funds and had to find employment quickly to make ends meet. He abandoned the venture.

In hindsight, he had not used the build measure-learn cycle.

Measuring and understanding feedback early on would have helped him make the necessary changes upfront. While going all-in brings great energy and drive to leaders and teams. Taking initial tentative curious but meaningful steps to learn and experiment saves organisations from costly mistakes.

Experiment cycle

Experimentation is at the heart of the culture and capacity building work we do with organisations globally. When looked at closely, the experimentation cycle mirrors the build-measure-learn approach. Experimentation involves framing a hypothesis - building a prototype and generating feedback. It fits the build-measure-learn cycle perfectly. Offering clues for organisations and leaders on how to implement the build-measure-learn cycle.

Ideas to implement build-measure-learn cycle

So, what can you as a business leader do to implement the build-measure-learn cycle in your team's ways of working? Here are two ideas to get started:

1. Possibility Thinking

When you have an idea on how a particular project/problem can be solved - Offer it as a possibility. Even better frame your idea as a hypothesis which can be tested. Hold the idea lightly. An accomplished HR head once told me- "Allow the idea to stand on its own merit and not because of who is supporting it"

2. Look for questions to answer

Another way to is to frame as a question - what do you need to learn about the project (that will help you choose better)? Then think of the least resource intensive way to start learning / measuring it in phases or sprints.

Agile stories

Now for the Zappos story we spoke of at the beginning. The founders of Zappos in 1999 opened an online store for shoes. Interestingly Zappos at that time did not manufacture or even sell shoes. The founders believed there was an untapped online market for shoes.

So they created a prototype website, and uploaded pics of shoes clicked from retail shops. They launched the prototype website and tracked buyer interest and orders. When the orders came in the founders went to the retail shops and bought the shoes and had them delivered!

The build-Measure-Learn cycle in action!

They wanted to learn if there was indeed a market for shoes online. They decided a good measure would be customer interest and final orders. Then they built a website prototype through which they could measure their criteria. Zappos continues to experiment its way to success even today.

How can you implement the build-measure-learn cycle in any aspect of your life? We invite you to consider it. To experiment with it.

Assumptions And The Ladder Of Inference

Deepa Premkumar — Tue, 21 Feb 2023 12:56:02 GMT

A discussion with my 9-year-old had me stumped. The discussion was ‘what protective gear was absolutely necessary while skating’. We decided to use data to make a decision. We both climbed to different conclusions using the same data! I wondered if i had missed a step in the decision-making process? Turns out decision making does involve steps and missing a step can be injurious! Let us explore these steps in decision making called as ‘The ladder of inferences’ in this post. This will help us become aware of our assumptions and better at decision making.

Origins of the ladder of inference

Chris Argyris, a Harvard professor created the ladder of inference. He was on an endeavor to discover ways of learning that produces change. Not superficial change but one that helps us explore our values and assumptions, to create sustainable change. Chris Argyris created the ladder when he was surprised that people looking at the same data were making different decisions. Much like, what protective gear to wear while skating discussion drew me to the ladder of inference.

“It's not easy to alter engrained behavior even when that behavior is clearly counterproductive — but, fortunately for the health of our organizations, it is possible”
Chris Argyris

Ladder of inference and skating gear.

Here is how it played out. We both could not agree on what protective gear was necessary and important while skating. We decided to watch videos of what professional skaters wore, to make our decision. We even selected and watched the same videos. So far so good. But it unraveled from there. The 9-year-old concluded that only the helmet would suffice. While i concluded that other elbow/palm guards were worn too and important!

How did we look at the same video(data) and draw different conclusions? Because, we selected different data points from what we were seeing and we interpreted it differently. We added our own needs and beliefs to the data to make a decision.

I added my need of safety to draw my conclusion that other protective gear was worn (in the video) and necessary. My daughter added her need to blend in (her friends were not wearing protective gear) and chose to go with one skater (data point) who was only wearing a helmet.

Debating our conclusions without being aware of how and why we made our decisions was futile. We see this scenario play out in many ways at the workplace.

Disagreements with the targets based on the same market trend report.
Different conclusions about the approach after a common debrief of the situation.
Walking out of an employee townhall with different interpretations about what was shared.

Why does this happen?

It turns out that we climb the ladder of inference so quickly, that we do not even realise we have. What is more some steps are so intuitive that we are not aware of having skipped them. These steps are closely linked to our beliefs and emotions, making it harder to spot and manage.

Even more surprisingly, we are surprised that people can people can interpret the same data differently. We assume that people around us share or must share our world view.

What makes skipping the steps in the ladder of inference and decisions that much riskier? It is the fact that the more times we skip the steps, the stronger our beliefs get. The stronger they get, the more selectively we chose data to support our beliefs. Talk of a vicious cycle.

How can we use the ladder of inference better?

Here are three ways to leverage the ladder of inference in your decision making

1. Retrace your steps in the ladder.

We climb the ladder very quickly. The approach to better decision making is to retrace your steps to the bottom of the ladder. Discussing or writing down the steps you followed to make the decision is helpful. Here are a few resources to do just that!

2. Compare assumptions and not conclusions.

Leaders often debate the conclusions that were drawn. A better use of time is to clarify what assumptions were made to draw the conclusions. This will help spot synergy and gap more easily, and lend itself to a constructive discussion.

3. Be curious.

Ask yourself often about what else might be true in the situation. Be open and curious to possibilities.

Awareness precedes change. By having an awareness of the ladder of inference we give ourselves an opportunity to make better decisions. All of us climb the ladder quickly, which is efficient but many times ineffective. Learning about our own assumptions and practicing going back to the data can improve all of our relationships and make us more effective leaders.

Agile OWL brings to you every fortnight an aspect of agile and experimentation. It explores the mindsets, skillsets and tools needed to be agile and adopt experimentation as a way of life. It also carries one story of experimenting from the world around us.

Agile Stories

Google turned many ‘this is how it is done’ myths on its head with this experiment.

When you need to choose colours for your website, you go to designers and not engineers. Even when you go to the designers, since it is an important page which will be seen by billions of people, the senior most designer’s opinion rules (HIPPO as they are not so affectionately called)

Google instead chose to run an experiment to test which shade of blue should be used on the ad-links. They collected data from these experiments and then chose the colour of blue!

Google identified 41 shades of blue that were available. Each shade was shown to 1% of the users. The number of clicks were studied. They repeated these experiments across demographics and analysed data. Finally identifying the top shade of blue which was deployed on their site.

Result? A 200 million dollar increase in ad-revenue from just the shade of the colour!

They continue their experiments 😊 Here are a few leaks.

References and Research

· Chris Argyris. Overcoming Organizational Defences, 1992

· Pete Senge. The 5th Discipline Fieldbook, 1994

· Ladder of inference Overview with question prompts

· The Executive mind and double loop learning by Chris Argyris

· Google and its 41 shades of blue experiment

Note : This post is Edition 51 of the Agile OWL from the OWL umbrella. The Agile OWL is a newsletter focused on the human experiences and stories within agile transformations. Sign up to receive the newsletter here

5 Tools To Spot Assumptions

Deepa Premkumar — Tue, 17 Jan 2023 04:35:58 GMT

We see the world through the lenses we use to see it from. Our lenses in turn, are shaped by our experiences, our beliefs and the assumptions we make. Thus, to see possibilities around us better, we need to be able to spot our assumptions. In this post we carry 5 tools to spot assumptions. Tools which will help us see assumptions and therefore possibilities better!

Assumptions & Agile

We have been exploring assumptions and its various facets in our earlier post(s). Assumptions are beliefs that we do not have evidence for. But assumptions play an important role in helping us work and take decisions. Agile ways of working help teams reduce the number of assumptions they need to make to take a decision. How do they do it? Teams take on a small area of work and approach it iteratively through experiments and sprints. Each experiment in each sprint increases what they know about the topic. Thereby reducing the assumptions, they have to make to take decisions.

“Your assumptions are your windows on the world. Scrub them off every once in a while, or the light won't come in.”
― Isaac Asimov

For Agile ways of working to be truly effective, teams must be able to spot and call out the assumptions they are making. Spotting and being aware of our assumptions is not always easy. Even experts are susceptible to these errors of spotting assumptions. Hence equipping ourselves with the skills, tools and mindsets required to spot assumptions is an important area for our own growth and development.

5 Tools To Spot Assumptions

Spotting assumptions helps bring awareness to the choices and decisions leaders and teams must make every day. In this edition we curate 5 tools business leaders can leverage to spot and call out assumptions at work and life. Read on to explore these tools and the links for more details.

1. Language we use

Yes, the language we use can be a good start point to identify assumptions we are making. We had written about it in detail here. You can take a quiz too to test yourself!

2. Customer Journey maps

Taking a step back to look at the customer journey's is a great way to call out some of the assumptions we are making about the product /process. Detailing out the steps, people involved, handovers etc, bring into focus many assumptions at play. Here is a resource to help you in creating journey maps.

3. Design Thinking

Yes, the concepts of desirability, viability and feasibility are great hunting grounds for assumptions at work. The secret here is to approach it with questions. Here is a resource which explains it in more detail.

4 & 5 Business Model Canvas & the Value proposition canvas

The business model and value proposition canvas are two powerful tools to explore the assumptions we are making. Writing out the pain points or gains from the product/feature/process for instance helps call out assumptions. The BMC on the other hand helps us become aware of the assumptions we are making about the macro environment too! These tools are even more useful together.

Peter Senge says it well, in the book Presence – “We all accept reality as it is presented to us. Our awareness presents itself to us as immediate and unmistakable. Yet there is always much more than we see.” Helping ourselves ‘see’ better makes the world one of possibilities and scenarios to explore. Let’s explore scenario thinking in our next edition.

Agile story

Here is a story of Team New Zealand and the America’s Cup. The world’s premier boating competition which began in 1851! Over the year’s the competition has become a high-profile international sports event that challenges not only teams sailing skills but also their ability to design, engineer and produce the most capable boats.

In 1995, Team New Zealand took on defending champions USA and won. So far Team New Zealand has won the title 4 times and are touted as the greatest modern Americas Cup team.

Team New Zealand credits their success to experimentation. Here is how they approached the race in 1995 and thereafter.

Team New Zealand led by Peter Blake agreed that their single objective was a boat that went faster. The team brain-stormed different design alternatives that could enhance the performance of the boat. Every idea generated to achieve this was treated as a different testable hypothesis.

Each hypothesis was then rigorously tested through a combination of simulated, prototyped, and live tests (Tank and tunnel tests)! This complementary approach of testing meant that Team New Zealand could manage time and budget constraints far better. They ran thought experiments, simulations, prototypes. Eliminating and building on hypothesis at every stage. Finally live testing those hypotheses that made the cut. They further generated data through live testing and implemented about 1/3rd of the ideas and hypothesis. In the process creating a sailing team which continuously learnt and improved and went on to win 4 championships!

Footnotes:

Agile Story:

McKinsey, Flying across the sea powered by AI.
John Childress, Will it make the boat go faster?, Association for Talent Development.
Toby Heppell, How New Zealand became the greatest modern America’s Cup team, Yachting World.
Marco Iansiti, Case study – Team New Zealand, Harvard Business Review.

Resources to spot assumptions:

Apollo M Nkwake, What are assumptions? Research on Springer Link.
Oschadleus, J. (2011). I think...I thought...what was I thinking?: Assumptions that drive decision-making. Paper presented at PMI® Global Congress 2011—North America, Dallas, TX. Newtown Square, PA: Project Management Institute.
Trafton Drew, Melissa L, Jeremy M Wolfe, The invisible gorilla strikes again: Sustained inattentional blindness in expert observers, Research Journal Sage publications (Vol 24, Issue 9)
Robert S Kaplan, Have you checked your assumptions lately? Inc.com and Harvard Business School.
Tool, Identifying assumptions and conclusions test, TopNotcher.
Tool, Business model canvas, Strategyzer.
Tool, Value Proposition Canvas , Strategyzer.

Assumptions Stop Us From Seeing

Deepa Premkumar — Wed, 14 Dec 2022 08:32:00 GMT

It has been an exhilarating football World Cup! And in the midst of all the madness, an on the pitch argument of whether a foul deserves a yellow card or a red got me thinking. Where does this debate of a factual event arise from? Each of us see reality from the lens of our expectations and assumptions, often without knowing that we do so. Where does this debate of a factual event arise from? Each of us see reality from the lens of our expectations and assumptions, often without knowing that we do so. Assumptions which we are not aware of / not called out, prevent us from really seeing and deeper learning. In this edition of the Agile OWL let us explore how assumptions stop us from seeing things through a story, a study and a quote. Well maybe two!

Assumptions and deeper learning

Deeper learning (as we explored in our earlier edition) occurs when we step out of our existing ways of thinking and are present to what is happening in front of us. What comes in the way of deeper learning is when we choose to react and re-enact habitual ways of thinking. When we continue to see the world through our assumptions, we lose an opportunity to learn and grow deeply and meaningfully.

How do we deepen our learning? By spotting assumptions and being able to test assumptions with experiments. Yes, spotting assumptions is hard and takes effort. Largely because assumptions are so insidious, implicit and ingrained at times. So ingrained that it prevents us from seeing things that are right in front of our eyes. We illustrate this point through a story, a study and a quote.

A story

This is a story from the early 1980's. Executives from US auto companies were wanting to find out how Japanese car manufacturers were outperforming them. Japan threw its doors open and welcomed US auto company executives to visit their plants. They were shown assembly lines and plants in operation.

Here is what the US auto company executives actually saw. The executives believed they were shown fake factories. Ones that were staged for this visit by the US competitors. Why did they think so? because the assembly lines and factories did not have stockpiles of inventory. Japanese were using the famous Just-In-Time production methodology.

The US car manufacturers were so ingrained with their expectations and assumptions of how a factory functions, that they did not see what was in front of them. What they saw was bounded by what they already knew. They had not developed the capacity to call out their assumptions.

In a few years when Japan overtook US as a leading car maker, it became painfully obvious that the factories they witnessed were indeed real.

A Study

This was a study conducted by MIT prof's Dan Ariely on how our expectations and assumptions impact our experiences. They set up several experiments in MIT to study expectations and experience of students. They had beer tasting experiments, coffee choices and many more.

In each of these experiments, participant choices differed when they already knew of the product. In other words when they had expectations and assumptions about the product, their experience of it varied!

Pepsi and coke found similar results when they did a blind study vs open eye study on which drink tasted better. When participants knew they were drinking Coke/Pepsi they rated the taste differently versus when they were blindfolded.

A quote

Here are two quotes to remind ourselves to be present to our inherent preferences, biases, and assumptions!

"You see what you know. You know what you see" – Frank Stella, Artist.

"It’s not what you dont know that gets you in trouble. It’s what you know for sure that isn’t so"- Mark Twain, Writer.

In conclusion, assumptions enable us to make sense of the world around us. Like we make sense of a conversation in a noisy room or patchy network despite the loss of a word here and there. But assumptions also come in the way of us seeing and being present to experiences around us. While stripping away our assumptions and previous knowledge is not possible. Perhaps we can at least acknowledge that we are trapped within our perspectives, which partially blinds us to the truth.

Agile Story

This is a story of one of the many experiments Kohl, the largest department store retail chain in America has conducted. Kohl believes in experimentation and data led decision making.

In 2013, Kohl was looking to reduce the operational costs of its stores. One of the ideas they came up with to reduce cost, was to open the store an hour later every day. This idea met with massive resistance from some members of the management team. The decision must be data driven they felt. The approach they adopted was to frame a hypothesis and collect data through experiments.

The hypothesis they framed was: "opening the stores an hour later in the morning from Monday to Saturday will reduce operational costs and not impact sales/revenue".

They identified 100 stores which were similar in all aspects. (Store size, neighbourhood demographics, sales, etc). These stores were split into two groups. In one set of stores, they changed the start time of the store to an hour later than usual. In the other set of stores, they kept the start time unchanged. The Kohl team measured operational costs and sales over a few months in both sets of stores.

At the end of the period, much to the surprise of veterans from the retail industry, data threw up interesting insights. Opening the store an hour earlier did not lead to any drop in revenues from the stores. What more, they were able to reduce operational costs in those stores too. Armed with this data store timings were changed across their 1158 stores across USA!

Work Trends OWL

Double-Loop Learning with AI: Getting Past the Why Barrier

When Why Becomes a Question About You

Single Loop with AI: Looks a Lot Like Success

Double Loop with AI: Breaks The Why Barrier

The Loop Runs Both Ways

AI & The Why Question

Value or Habit? What AI Is Forcing Us to Ask

Value or Habit?

Value or past success?

Value or comfort zone?

AI Forces Us To Ask ‘Why’!

Noticing Our Emotions With AI

Valuing capacity building with AI

How Is Identity Built In The Age Of AI

The Unease Professionals Are Carrying

Identity Is Curated, Not Delivered

Experiment Your Way to AI

Further Reading and References

The foundational thinkers

Additional Research

AI And The Case Of Mistaken Identity

The Question Nobody Warned You About

Why Tasks & Execution?

Not the Same Thing

The Reframe AI is helping us with

Why Is Generative AI Asking An Identity Question

Why does AI feel different?

One, AI speaks your language.

Two, Speed of AI Creation

Three, It removes the skill excuse.

The Real Question

Identity Recrafting in the Age of AI

What am I letting go?

What am I taking on that will add value?

If this technology disappeared tomorrow, would I be inconvenienced or handicapped?

Further Reading

On pilots and the cost of over-relying on automation

On radiologists actively recrafting their identity in response to AI

On identity shifts during professional and career transitions

On the psychology of letting go during change

AI is asking an Identity Question Of us

The GPS and the London Cab Driver

The Camera and the Portrait Painter

The Industrial Revolution and the Craftsman’s Pride

AI Is Asking the Same Question — Again

The Software Stopped Waiting for You

What actually happened?

The people who saw it coming

The question nobody is asking out loud

Every big shift has done this

So what do you actually do?

Organisational Singularity: Explained Five Ways

To a 5 Year Old

What you can try now to be ready for Singularity:

To a College Student

What you can try now to be ready for Singularity:

To a New Hire

What you can try now to be ready for Singularity:

To an L&D Head

What you can try now to be ready for Singularity:

To a CEO or Senior Leader

What you can try now to be ready for Singularity:

In closing

Organisational Singularity: What It Is and Why It Matters

What Is Organisational Singularity?

Why is organisational singularity becoming possible now?

Principles of Organisational Singularity

The Compounding Principle

The Systems Principle

Evolution (Not Extinction Principle)

Organisational Singularity is Underway Already

Further Reading & References:

From Execution to Insight: The Shift AI Is Making Possible

Two Ends of a Spectrum

Why Execution Pulls Ahead

What AI Is Changing — And What It Is Really Asking

Recrafting the Role

Recrafting the Organisation

An Invitation