One of the most unnerving decisions is what AI tools should you adopt and utilise. With so few scaled operational case studies it’s a risky business.

It may all be new but we can ask sensible questions to form a more rounded opinion.

Your selection challenge

There is more to consider than it’s cool or a consultant recommended it.

Do you know what the AI Tool actually does?

I came across this maturity framework and refined it to assess a client’s AI use case portfolio. I found it very useful in categorising the portfolio of use cases and then aligning the best fit AI tools to each AI category.

Read more

In 2018 I joined Unilever and inherited the beginning of an ambitious supply chain data lake build. The technical architecture had already been set and the data teams were busy building out the extraction and transformation layers. But from that day till four years later we could never effectively describe to sponsors, leaders and product owners what we were building and what the capabilities would be.

It was just referred to from the board to the individual users as the data lake that would supply data for all applications and use cases.

There was a lot of frustration and disappointment as the data lake didn’t provide the data in the time sequence or granularity that product owners needed.

This all stemmed from not being able to communicate what it was designed to do and what it could provide.

In this AI revolution you may be a consultant, tech vendor, client buyer or sponsor who wants to deploy AI but know you have to organise your supply chain data in a AI ready data platform. If that…

Read more

Because traditional databases were costly to design and operate, their scope stayed limited. The result is a fragmented legacy landscape owned by IT and stitched together by spreadsheets.

To overcome this scenario enterprises have been building cloud based enterprise data lakes for nearly 10 years but are they ready for AI applications?

Here are 5 challenges you face that mean your existing data platforms my not be fit for AI uses cases.

1. Do you understand what your data means? - Creating a data platform that is the base of many AI us…

Read more

In the last 12 months I have been working on a number of projects in the AI and supply chain space so have new and recent content to share.

I feel ready to reignite the Substack and will start posting again once or twice a month starting next week.

The new content will be free for 2 weeks, with the historic posts behind a paywall.

I hope you find the new content insightful and useful.

Read more

That country is the USA. It can do this because all major cloud infrastructure are owned and controlled by US companies. You may think not a problem as IT ecosystems are diversified but large organisations are building data platforms to consolidate, transform and enable data for AI applications on these USA controlled cloud infrastructures. Rather than having 20+ diversified enterprise systems organisations are now creating one data platform.

That is a single point of failure.

AI will concentrate this single point of failure as AI applications replace humans, create enhanced processes that only AI can handle and create automations that no human ever touches. Supply chain offers some of the biggest AI opportunities which means operational supply chains…

Read more

There is a lot going on in the world that will transform how global supply chain’s operate. As this Substack is about digital supply chain there has been one threat that is openly stated but I haven’t seen being picked up by anyone.

My thinking started on this during a mentor call I had with an AWS employee who was concerned about the vulnerability of the regular use of open source software in supply chain solutions. Small open source solutions often supported just by the original developer can be used within a bigger solution, leaving the whole solution at risk of failure if the open source software is not supported and maintained effectively.

I thought that was interesting and worth keeping an eye on but was isolated in it’s impact and was manageable with good IT governance.

I spend a lot of time training outside and in the gym so listen to Podcasts. I listened to a Joe Rogan Podcast with Mark Zuckerberg as you don’t get to hear a business leader like him talk for 3 hours.

It was recorded at the beginning of January 2025, just after he announced a change to the content moderation process at Facebook. He explained his logic for why human moderation was flawed and how the AI moderation penalised content creators when the algorithm wrongly identifies them.

His argument was that free speech is better and let the community challenge the facts.

Why was he doing this? His reason was countries outside of the USA were creating laws to make him and his company financially and in some cases criminally liable if content they deem dangerous to their citizens was published on social media platforms.

The next statement from him made me sit up. He said the US government should treat it’s tech firms as a strategic asset and should defend them from other countries stopping them trade as they wanted in those countries.

This was a quid pro quo. I will stop the moderation of content that Republicans want broadcast and in return you will protect me from being attacked by these countries.

Read more

If the goal is for a digital supply to operate with humans out of the loop, the optimisation decisions normally made by humans will need to be done by AI algorithms.

But when we say optimisation in supply chain we think about network or inventory optimisation decisions around strategic planning, using periodic reviews rather than real time operational optimisation.

A key difference of a digital supply chain is a sense and response capability to replace the present, analogue plan and hope model. The present model uses humans in the loop on most operational decisions, who create a risk adjusted plan that they hope covers all the scenarios of demand and supply fluctuations they might face in the future. They do this as humans can’t compute and assess complex, large data in real time.

Therefore to create real time sense and respond capability within complex systems that make optimal decisions, optimisation algorithms will be required that will take humans out of the loop, and process large…

Read more

Here is the data problem for AI in enterprises. AI needs big and complete data to be trained on, but if all data was stored the cost would be so great it would wipe out any AI business case e.g. one Tesla car will generate approximately 100 - 250TB of data per year which at 15 cents per GB per month (Azure pay as you go premium blob storage) would cost $460,800 to store a year.

The result is typically only 3% of data generated gets to a data centre due to cost.

This means in a supply chain with an increasing amount of data being generated by devices and vehicles in localities, this cost data problem will become prohibitive for business cases to get approved.

To solve this business case cost problem, AI models will have to move from being stored and run in data centres, to the edge where the data is generated.

Moving to the edge allows the AI algorithm to access the data as it is generated, to extract the elements it needs and turn it into signals that can be passed back to the centre as …

Read more

This is a bonus post as a follow on to #15 Gen AI - Why All the Hype in Supply Chain? Click here to read it.

In that post I introduced the concept of Gen AI Best-Self coaches who have access to all the data available and can provide specific answers and insights in real time to people managing supply chains.

I was at an event last week at the National Innovation Centre for Data in the UK and listened to a keynote talk by Bill Thompson who leads BBC R&D as well as being an author and podcaster.

He presented an argument, as an emerging idea that was still “a bit flaky”, that Generative AI (Gen AI) and Large Language Models (LLMs), could close the circle of humans having access to all the written word content available in the world, which could result in a social and economic revolution.

The argument goes like this. The evolution of communicating information via language has gone through four stages:

1. Human sharing of information via stories, songs, fables, myths, prayer. Confined to local gr…

Read more

So this one will introduce the concept of Reinforcement Learning which I believe will become an important AI optimisation technique in supply chain planning.

For the uninitiated, Fight Club is a 1999 American film about a club with 8 rules.

1. You do not talk about Fight Club.

2. You do NOT talk about Fight Club.

3. If someone says "Stop" or goes limp, taps out, the fight is over.

4. Only two guys to a fight.

5. One fight at a time.

6. No shirts, no shoes, no mercy.

7. Fights will go on as long as they have to, even if it’s over a month.

8. If this is your first time at Fight Club, you have to fight.

When I joined Unilever in 2018 a Fight Club between planners and machine learning demand forecasting algorithms was established to create a competition around forecast accuracy that would result in planners being taken out of the loop. I’m therefore breaking Rule 1 and 2 with my post.

The simple challenge I was given, was can your machine learning algorithm beat a planner. It was reinforced by a board member who said if yo…

Read more

I’m back!

This is the first post in a series on AI in supply chain, covering what I believe will be the models and capabilities that will have the biggest impact on digital supply chain.

AI has blown up as a hype cycle in a way I haven’t seen since the Y2K hype when the world was predicted to end if all IT systems weren’t upgraded (you may not have even been born then, so Google it). The world didn’t end but consultancies and IT firms made a lot of money out of it.

The last revenue generating crisis ended a year ago i.e. Covid and all the investments in services and technology that were made to manage the crisis e.g. according to the National Audit Office the UK Government spent £25.7 billion by June 2022 on just Test and Trace.

The service industry needs a new crisis reaction to break through the buying barriers of organisations they sell to, especially when budgets are tight due to inflationary pressure. AI and specifically Generative AI (Gen AI) seems to be at the forefront of that.

I…

Read more

I asked ChatGPT what are the best use cases in supply chain for AI and this was the answer. Which I think was pretty good.

The first post in the series will cover use case #15, actually looking at Gen AI co-bots rather than Chatbots. This is an easy entry point into the topic area and I will ramp up the complexity as we progress through some of the other use cases during the series.

ChatGPT Answer

“Artificial Intelligence (AI) offers several valuable use cases for optimizing and improving supply chain operations. Here are some good AI use cases for supply chain management:

1. Demand Forecasting: AI can analyze historical data, market trends, and external factors to provide accurate demand forecasts. This helps companies optimize inventory levels, production schedules, and distribution strategies, minimizing stockouts and excess inventory.

2. Inventory Management: AI can determin…

Read more

Back after a summer break.

Having spent 14 years trying to persuade people that advanced analytics is the future it has been nice to see so much focus on AI in the last 6 months.

The AI term is used very generically by most people, so the next series of posts will go deep into areas of AI that are likely to be the future of Digital Supply Chain.

They will start at the beginning of September and come out once a month.

I’m also launching a new Substack called Longevity Experience focused on my Longevity health and fitness practice that I have been setting up over the last year. Sharing content and experiences of helping busy people aged 35 to 65 years old improve their health so they can perform better in work and life.

Please subscribe to that one if you are interested in your health and longevity.

The Longevity Experience

A newsletter providing practical experience on fitness, physiological and emotional health for 35-65 year olds, who want to focus on their longevity, so they can perform, feel and look better as they age to lead a healthier and more fulfilling life.

By Nigel Issa

Read more

1. Digital Supply Chain in Practice

Mars Wrigley’s Highly Successful Supply Chain Digital Transformation

Click here to find out more

Read more

In previous posts I have touched on a key future capability.

Leaders will need to be able to manage both humans and machines in their teams. By machines I mean algorithms and models that make decisions rather than humans. In this post I’m going to uncover what that capability could look like.

If you are already thinking not another post about ChatGPT or AI don’t worry those words will not be mentioned again in this post. If you want to explore the ethics around these two things, I suggest you look at what two ex-colleagues of mine, Ray Eitel-Porter and Giles Pavey have written on this subject, going back many years.

This will be a practioner view, about operationally using machines in supply chain.

You may be thinking what supply chain decisions are we talking about?

These are the ones I think have the most probability of humans being out of the loop whem supply chain decisions are made. At the moment humans are generally in the loop with technology and data provided to improve or augme…

Read more

1. Digital Supply Chain in Practice

Logistics providers bet on tech to solve supply chain problems

Click here to find out more

Read more

It’s been suggested I have neglected procurement so far on this Substack, so this post will address that and try to shed some light on why, despite a large investment in digital procurement capabilities procurement teams still hope spend goes right rather than proactively managing cost, reliability and resilience using the capabilities they often have access to.

Procurement has always had some of the best data and most structured processes across all the supply chain functions. For a long time they have had access to a vast array of digital solutions that can identify and capture value and risk reduction opportunities.

But despite this, procurement organisations don’t seem to change from the deal making culture of periodic reviews to managing everything all of the time that digital solutions can enable.

What’s holding them back? As the saying goes you can lead a horse to water…

Read more

1. Digital Supply Chain in Practice

Delivery firm Gently pursues growth with nano-fulfillment centers

Click here to find out more

Read more

When I started this Substack one of the themes was going to be analytics @ scale and this post was going to be about why you need an army to deliver analytics @ scale.

Having had time to reflect and observed how scale has gone wrong and created complexity, high costs and longer timescales this is going to be a story of what not to do.

Why do I say it is a problem?

There is a contradiction between an analytics delivery and support model that is very labour intensive, but at the same time the desire and expectations to move quickly and be agile as it is an area where sponsors, users and delivery people are not very clear on what to do, requiring many direction changes.

This creates the worst case scenario of being costly, difficult to manoeuvre and slow to deliver.

For years employees at the big tech firms like Google and Facebook reported how many managers there were and how limited their roles were leading to large proportion of working time spent coordinating between each other rather …

Read more