AI Adoption Framework

A framework for leaders who don't code, written for the leaders who have to make AI decisions anyway.

Why this exists

Most AI strategy content is written for technical leaders, by technical leaders. That's a problem, because most AI decisions are not technical decisions.

This Canvas is for you if you are:

• A CEO, CFO, COO, or senior leader without a technical background

• Responsible for AI decisions you do not fully understand technically

• Trying to translate AI capability into specific business value

• Looking for a map to know where you are and what comes next

What you will get from this page:

• A clear view of where your organisation sits on the AI adoption journey

• The critical decision at your current stage

• The most common trap at your current stage

• A practical signal that confirms which stage you are actually in

The journey at a glance

There are five stages. Most organisations are stuck between Stage 2 and Stage 3.

The Framework: five stages, five decisions, five traps

Stage 1: Exploration

What it looks like:

• Leadership has heard about AI in board conversations

• A few pilots are running, often informally

• No coherent strategy, no budget line, no governance framework

• AI is a personal interest of one or two leaders, not yet an organisational priority

Who fits here:

• Companies in the first 6-12 months of taking AI seriously

• Mid-market businesses that have not committed budget

• Founder-led organisations where AI is curiosity, not commitment

The critical decision: Which workflows are worth the first real investment?

Not which tools. Not which vendors. Which workflows. The decision is about where AI gets applied first, because the first deployment teaches you everything you need to know to make the second one work.

The common trap: Spreading thin.

• Running fifteen small pilots produces fifteen disconnected learnings

• None of them add up to organisational capability

• Teams that win run one or two real pilots with real measurement attached

The honest signal you are here: You cannot answer "what is our AI strategy?" with anything more specific than "we are exploring AI."

Stage 2: Deployment

What it looks like:

• Copilot is licensed across the organisation

• Maybe Claude or other tools too

• Specific teams use AI daily

• Productivity is measurably up

• The conversation has shifted from "should we?" to "how do we do this better?"

Who fits here:

• Most knowledge-economy businesses with 200+ employees in 2026

• Companies past the first wave of deployment

• Leadership teams planning the second wave

The critical decision: How do we measure value beyond adoption?

Adoption metrics measure motion. Outcome metrics measure progress. Most leaders measure the wrong one.

The common trap: Adoption theatre.

• Leadership celebrates "we deployed AI to 5,000 seats" without anyone asking what happened to the work those 5,000 people produce

• Six months later, productivity numbers plateau

• The "transformation" is invisible on the P&L

The honest signal you are here: You can name the tools you have deployed. You cannot name the business outcomes those tools have produced.

Stage 3: Integration

What it looks like:

• AI is no longer a tool layer sitting on top of the business

• It is embedded in specific workflows

• Sales proposals are AI-drafted and human-refined

• Customer support is AI-triaged and human-resolved

• Specific cycle times are measurably down

• The CFO is paying attention

Who fits here:

• Companies 18+ months into serious AI work

• Organisations that have invested in workflow redesign for at least one critical process

• Teams seeing real ROI on specific use cases

The critical decision: How do we govern this without slowing it down?

Governance done well must operate at the speed of AI deployment. Done badly, it either bottlenecks the work or rubber-stamps the work.

The principles that separate real governance from theatre:

• Tiered review: lightweight for low-risk agents, deep for high-risk

• Named accountable owners, not committees

• Measure outcomes, not activity

• Insert governance into procurement, not parallel to it

The common trap: Theatre governance.

• Policies are written

• Committees are formed

• Quarterly reports are produced

• Nothing actually gets stopped, reversed, or denied

• The framework looks good and does not catch anything

The honest signal you are here: You can name at least two workflows where AI is producing measurable value. You also have at least one nagging concern about an AI deployment you do not fully control.

Stage 4: Redesign

What it looks like:

• The conversation has shifted from "how do we use AI?" to "what would this work look like if AI were the primary actor?"

• Roles are being rewritten

• Reporting structures are being questioned

• New job descriptions emphasise judgment over execution

• Some teams are smaller than 18 months ago

• Some teams are doing work that did not exist 18 months ago

Who fits here:

• A small minority of organisations

• The leading edge of enterprise AI adoption

• The companies you read about in case studies, not the ones writing them

The critical decision: What do we restructure first?

Linear AI use produces linear gains. Compounding gains require structural change.

The companies that figure this out in 2026 will be uncatchable by 2028.

Operating model redesign is sequential:

• Try to do everything at once and you stall

• Try to do nothing and you stay in Stage 3 indefinitely

• The right approach: pick one core workflow, redesign it from scratch, use the proof point to fund the next

The common trap: All-at-once redesign.

• Leadership commits to a comprehensive transformation programme

• 18 months later, the programme has produced PowerPoint decks and consulting fees

• It has not produced operational change

The honest signal you are here: You have at least one job description rewritten because of AI, and at least one decision authority that has moved closer to where AI is being used.

Stage 5: Compounding

What it looks like:

• AI is not a tool, a strategy, or an initiative

• It is an operating assumption

• The business is organised so every use of AI makes the next use of AI better

• New capabilities emerge that were not on the original roadmap

• The team is doing work that was unimaginable three years ago

Who fits here:

• Almost nobody, yet

• Maybe a handful of frontier technology companies

• The companies who reach this stage in the next 24 months will be uncatchable by 2028

The critical decision: How do we keep accelerating without burning out the team?

Compounding gains create their own pressures:

• Teams that produce more with less also feel more, judge more, and decide more

• The cognitive load is different but not lower

• The companies that sustain Stage 5 protect human capacity as carefully as they protect AI capacity

The common trap: Productivity gains that erode meaning faster than the organisation can adapt.

• Cycle times drop

• Output rises

• Adoption is high

• The best people quietly leave because the work no longer feels like theirs

The honest signal you are here: You can name a new capability your organisation has now that did not exist three years ago, and that capability does not depend on any specific AI tool.

How to use this Canvas

Three steps. Do them in order.

Step 1: Locate yourself.

• Read each stage carefully

• Identify the stage that most accurately describes your organisation today

• Most leaders place themselves one stage too far along

Step 2: Confront the critical decision.

• Each stage has one critical decision

• Yours is the one you have been deferring, not avoiding

• Bring it to your next leadership conversation

Step 3: Avoid the trap.

• Each stage has one common failure mode

• Yours is the one your organisation is currently exhibiting

• Naming it is the first move toward addressing it

Most leadership teams I work with discover three things when they apply this honestly:

• They are not as far along as their internal narrative suggests

• The critical decision at their stage is one they have been deferring, not avoiding

• The trap at their stage is exactly the failure mode they are currently exhibiting

If any of those three sound familiar, you are ready for the next conversation.

What this Canvas does not do

This framework is a map. It is not a plan.

• It tells you where you are

• It tells you what comes next

• It does not tell you how to get there in your specific business, with your specific people, your specific data, and your specific competitive context

That is the work I do with clients. If the Canvas resonates and you want help moving from one stage to the next, the conversation starts with a 30-minute call.