An AI- Enabled Organisation, or an Organisation With AI in It?
The AI Vortex is a 4-part series on how organisations move fast on AI and end up short on return, and what the organisations getting it right do differently.
Part 1, you can read it here, covered the pull of the possibilities and the cost of an AI roadmap that is incomplete, sits in pockets of the business, and cannot absorb new capability. This part covers the 2 decisions that make an AI roadmap complete.
I see the gap most clearly in executive briefings where an AI program is presented for funding. The presentation covers the tool, the vendor, the use cases and the timeline. It rarely covers the 2 decisions that decide whether the program will still be useful in 18 months:
The AI Role: how much of the work the people in the organisation keep, and how much AI does. The 4 AI Roles are assistant, copilot, doer and autonomous.
The AI Usage Category: what the organisation wants AI to do for the business. The 5 AI Usage Categories are the Amplifier, the Edge Maker, the Capacity Creator, the Wiz and the Housekeeper.
Deciding what kind of AI purpose you want, and how you are going to use AI, makes a huge difference to the design, the structure, the talent you need, the budget, the governance, and the continuum of where you are going to build AI and continue to evolve it. Skip the AI Role and the AI Usage Category decisions and the technology makes them for you, one release at a time.
TL;DR
The first decision is the AI Role: assistant, copilot, doer or autonomous. It sets how much of the work the people in the organisation keep and how much AI does, and it is decided per team and per kind of work rather than once for the whole organisation.
The second decision is the AI Usage Category: the Amplifier, the Edge Maker, the Capacity Creator, the Wiz or the Housekeeper. Overlaps exist, and the decision is which category you lead with.
The AI Role and the AI Usage Category together set the design, the structure, the talent, the budget, the governance and the continuum. The same AI tool in a different AI Role, or for a different AI Usage Category, is a different program.
AI treated as a project with an end date drifts. The organisation keeps paying for it while it becomes less useful, and calls itself AI-enabled while using 5% to 10% of what AI can do for it.
The research distinguishes between the organisations that make these 2 decisions and those that do not. McKinsey's 2025 survey found that AI high performers are nearly 3 times as likely as other organisations to redesign their workflows when deploying AI.
🎧 I highly recommend you to listen to this audio podcast as a deep dive which is AI-generated based on this edition and Jess' intellectual property:
https://gotoleader.s.gy/an-ai-enabled-organisation
The first decision: the AI Role, and which of the 4 you want for the people in your organisation
The AI Role is decided for the people in the organisation, and it comes before any decision about the technology.
The question I ask an executive team before any use case or vendor is which AI Role they want for the people in their organisation. I have asked it on programs from 1 million to 350 million dollars, in organisations of 1,000 to more than 10,000 people, and the answer has changed the design each time.
The AI Role is a decision about appetite, trust and risk, and executives can make it today from what they know about their own organisation. There are 4 AI Roles that describe how much of the work shifts from people in the organisation to AI.
Assistant. AI sits alongside the person, and the person stays in control of the work. AI drafts, summarises, retrieves and suggests, and the person decides what happens with it.
Copilot, in the plain sense of the word. The person directs the work, observes AI performing defined parts of it, and remains present throughout.
Doer. The person offloads tasks to AI, sets each task, reviews the output, and is no longer present while the work is done.
Autonomous. The organisation offloads much more of the work to AI, including the decision about what to do next inside agreed limits. The person sets the limits and reviews the results.
The AI Role is chosen for a team and a type of work, and one organisation can carry several AI Roles at once. A finance team can run reconciliations with AI as a doer, while the strategy team in the same organisation runs analysis with AI as a copilot. Setting one AI Role for the whole organisation results in the one-size-fits-all implementation I described in Part 1, which I see as playing too small.
Each step from assistant toward autonomous moves risk, skills and governance with it. With AI as assistant, the person catches errors as part of the work. With AI as a doer or autonomous, the organisation needs a way to catch errors that does not depend on someone reading each output, and that is a design and governance decision rather than a training one.
The research supports treating the AI Role as a deliberate decision. McKinsey's 2025 State of AI survey found that 62% of organisations are experimenting with AI agents and 23% are scaling them in at least 1 function. The organisations that reach scale have decided how much work to move to AI, which teams to move to AI, and what must be true first.
📍 When done right, this is what it looks like
The AI Role is written into the business case, per team and per kind of work, so the governance and skills requirements are visible before spend.
The plan specifies which teams move from one AI Role to the next and what must be true first, based on the organisation's own maturity.
The people in each team know which AI Role applies to their work because it tells them what is expected of them and what remains theirs.
📍 In Practice
Take your 3 largest AI initiatives and write the AI Role against each one: assistant, copilot, doer or autonomous. If the same word applies to all 3, ask whether it was chosen or defaulted.
For any initiative with AI as doer or autonomous, write down who reviews the output and how often. If no reviewer has been assigned yet, that initiative is running in a bigger AI Role than it was designed for.
Ask each business unit head which AI Role they want for their unit's people over the next 12 months. The spread of answers marks the start of an AI roadmap integrated across the business.
The second decision: the AI Usage Category, and which of the 5 leads
Each new AI capability arrives with a ready answer to what AI should do for your organisation, and the AI Usage Category is your own answer.
Once the AI Role is decided, the second decision is which AI Usage Category the organisation leads with, and this is the decision the vortex makes for an organisation that leaves it open. Each new AI capability arrives with a ready answer to what AI should do for the business, and an organisation without its own answer takes whichever one arrived most recently.
I use 5 AI Usage Categories for this decision. They come from the programs I have run across specialised deployments within a single function and enterprise portfolios running in parallel, and each one has changed the design.
The Amplifier, AI and human potential. AI is used so the people in the organisation do more of what they do best, and the measure is the quality and reach of their work.
The Edge Maker, AI as competitive advantage. AI is used to do something the market cannot easily copy, and the measure is market position.
The Capacity Creator, AI to scale the organisation. AI is built in 1 part of the business first, then compounded across the rest, and the measure is what the organisation can now take on.
The Wiz, AI as a thinking partner. AI reads from multiple sources and raises the level of thinking, and the measure is the quality of the decisions that follow.
The Housekeeper, AI as organiser and reporter. AI keeps the organisation in order, and the measure is time and accuracy. The Housekeeper is the most common use I see today.
The 5 AI Usage Categories overlap, and most organisations use more than one. The decision is which category you lead with, because the lead category is what the design, talent, and governance get built around.
The AI Usage Category also combines with the AI Role. An autonomous Housekeeper is a low-risk decision with a clear governance model, and an autonomous Edge Maker is a strategic decision that reaches the board. Same AI Role, different AI Usage Category, and a different program in each part that costs money.
"We do not need a category, we need results." I hear this from sponsors, and the results are the point. The AI Usage Category is how the results get defined. Without it, the program reports whatever results the tool happens to produce, which is how a program ends up reporting time saved when the business case was written for competitive position.
BCG's 2025 study of 1,250 companies found that 70% of AI's potential value sits in core business functions rather than in support functions. That is a finding about the AI Usage Category, because the Housekeeper is the most common lead category and it sits mostly in support functions.
📍 When done right, this is what it looks like
The lead AI Usage Category is written into the business case, in the organisation's own words, and the measures follow from it.
The lead AI Usage Category sets the talent plan and the governance model, so the organisation is not recruiting for the Housekeeper while promising the board the Edge Maker.
The AI Role and the lead AI Usage Category sit on the same line in the business case because together they provide the shortest, most accurate description of the program.
📍 In Practice
Ask the sponsor of each AI initiative which of the 5 AI Usage Categories it leads with, in 1 sentence: the Amplifier, the Edge Maker, the Capacity Creator, the Wiz or the Housekeeper. If the answer is a tool name, the category has not been decided.
Compare the lead AI Usage Category to the measures the initiative reports on. If a program that claims the Edge Maker is reporting hours saved, the measures belong to a different category.
What the AI Role and the AI Usage Category change: design, structure, talent, budget, governance and continuum
The same AI tool in a different AI Role or for a different AI Usage Category. The choices made create vastly different work programs, with different sets of expectations and budgets.
The AI Role and the AI Usage Category together set the design, structure, talent you need, budget, governance, and the continuum of where you will build AI and continue to evolve it. Each of those 6 is a line in the program, and each one changes when the 2 decisions change.
Design. An Amplifier with AI as assistant is designed around the person's workflow. A Capacity Creator with AI, when autonomous, is designed around the process, and the person's job is redesigned around the AI. McKinsey's 2025 survey found that AI high performers are nearly 3 times as likely as other organisations to fundamentally redesign their workflows when they deploy AI.
Structure. The AI Role decides who is accountable for the output. With AI as autonomous, someone has to own the results of work no person watched, and that job often does not exist yet.
Talent. The Wiz needs people who can direct and challenge AI thinking, the Housekeeper needs people who can specify and check, and the Edge Maker needs people close to the market. Recruiting for the wrong AI Usage Category is expensive, because the people arrive after the tools are bought.
Budget. The AI Usage Category decides where the money goes after the licence. The Capacity Creator spends on integration and the compounding path, and the Amplifier spends on workforce enablement. A budget that is mostly technology is a budget that assumes the Housekeeper, whatever the business case said.
Governance. The AI Role decides how much governance is needed, and the AI Usage Category decides what the governance protects. A Housekeeper with AI as doer needs accuracy checks, and an Edge Maker with AI as autonomous needs a governance model the board understands and has agreed to.
Continuum. The continuum is where the organisation is building AI now, where it intends to build next, and how it will keep evolving what it has built. The AI Role and the AI Usage Category describe the starting point, and the next section covers what happens when the organisation stops at that point.
The organisations I have seen recover from a poor first build could describe the AI Role and the AI Usage Category they had built, compare them to what the business needed, and see the gap. The organisations that struggled had a tool, a vendor and a set of use cases, and no shared description of what the program was for.
📍 When done right, this is what it looks like
Design, structure, talent, budget, governance, and continuum are each documented against the AI Role and the AI Usage Category before the program is funded, and the business case explains how each was decided.
The talent plan is written before the tool is purchased, and the budget split between technology and enablement aligns with the AI Usage Category.
📍 In Practice
Take the business case for your largest AI initiative and find the 6 lines: design, structure, talent, budget, governance and continuum. Mark which are written against a stated AI Role and AI Usage Category, and which are written against a tool.
Ask who is accountable for the initiative's output where AI plays the biggest role. If the answer is a system name, the structure decision has not been made.
Treat AI as a project with an end date, and it drifts
An AI implementation with an end date keeps costing money after the end date, and becomes less useful every quarter.
If you treat AI as a project with an end date, the implementation drifts and the return goes negative. The business changes, the data changes, and people's behaviour, mindset and the way they perceive things change. An AI implementation designed for one kind of work at one point in time becomes less useful each quarter it is not evolved, while the licences, maintenance, and running costs continue.
This is the "set and forget" pattern, and I see it in well-implemented programs.
The continuum is the answer. It means the organisation has decided where it is building AI now and where it builds next. It also means the organisation has decided how it will move teams from one AI Role to the next and the business from one AI Usage Category to the next over time.
And it means the organisation has decided how it keeps what it has built governed, up to date, and checked, so that it does not diverge or drift from what it was built for.
The evolution runs both ways. The organisation asks how AI can help it evolve, and it asks how it can help AI evolve by feeding AI better data, better limits and better questions as the organisation learns what it wants. An organisation that does the first without the second has bought a capability and is running it as an add-on.
Treating AI as a project with an end date is also where the box gets ticked. An organisation with the Housekeeper as its AI Usage Category and AI as assistant in 3 functions can report to its board that it is AI-enabled, and the statement is accurate. It is using 5% to 10% of what AI can do for it, and nothing in the program was designed to move.
📍 When done right, this is what it looks like
The program has the next AI Role and the next AI Usage Category written down, along with what must be true before the organisation moves to them.
The implementation is reviewed on a set cycle against its intended use, and the organisation has a way to feed AI what it has learned.
📍 In Practice
For each live AI initiative, find the date it was last reviewed against its original purpose. Anything over 12 months is running on assumption.
Write the next AI Role and the next AI Usage Category for your lead initiative, and the 1 condition that has to be true before you move. If you cannot write the condition, the continuum has not been designed.
The takeaway
The vortex makes the AI Role and the AI Usage Category decisions for an organisation, leaving them open and basing them on whatever capability arrived most recently.
Deciding the AI Role (assistant, copilot, doer, or autonomous) and the lead AI Usage Category (Amplifier, Edge Maker, Capacity Creator, Wiz, or Housekeeper) yourself is what puts the AI roadmap back in the hands of the business.
Those 2 decisions set the design, structure, talent, budget, governance, and continuum. Change either decision and the program changes in each of those 6.
AI treated as a project with an end date drifts, and the organisation keeps paying while the return falls.
The leaders who can describe their organisation's AI program as an AI Role and an AI Usage Category, and say where it goes next, are the leaders executives bring in when the next capability arrives. That description is the difference between an AI-enabled organisation and an organisation with AI.
Write the AI Role and the lead AI Usage Category for your largest AI initiative on 1 line. If it took more than a minute, which decision was missing?
See where your AI transformation stands across 10 areas
The AI Transformation Leader Scorecard assesses you across AI strategy, proof-of-concept design, governance design, stakeholder navigation, vendor navigation, adoption and workforce, your own AI transformation skill set, your professional positioning, and AI maturity.
It gives you a scored read on each area, which is the fastest way to see whether the AI Role and AI Usage Category decisions in this piece have been made in your organisation or defaulted.
https://www.futurefitleader.com/ai-transformation-leader
My invitation to you: go deeper into AI transformation, so you are equipped with the right thinking and tools
The AI Transformation Masterclass is the practical version of what 5 years of being embedded in AI transformation have taught me.
It covers what AI is and is not, and the use cases that hold up over time. It also covers how to be in the AI conversation in your business as the person who can describe the program, rather than the person who was handed the tool.
https://www.tl-institute.com/c/ai-transformation
Next in the series: Part 3 asks who decided that more productivity was the aim, and what the people in your organisation hear when it is.
