Quick Navigation
- The cost of getting it wrong
- AI vision — where is AI taking you?
- Operating model — how to organise for AI
- Maturity assessment — where are you today?
- Value framework — how to measure AI impact
- Strategic portfolio — what to fund first
- Building the roadmap
- The strategy stack in one diagram
- Cheat sheet
The cost of getting it wrong
Most organisations fail at AI not because the technology doesn't work, but because they never decided what it was for.
AI without strategy is expensive experimentation. Organisations that treat AI as a collection of disconnected pilot projects end up with pockets of value, duplicated effort, and no compounding advantage. Every team picks its own tools. Pilots never connect to business goals. Governance is ad hoc or absent. Talent is scattered. Leadership sees cost, not value.
The failure pattern is predictable: organisation sees AI hype, multiple teams start pilots, each pilot works in isolation, the same problems get solved multiple times, pilots can't scale, leadership loses patience, the AI programme gets paused or abandoned.
The cost is not just wasted money. It is wasted time, lost credibility, and an organisation that becomes harder to transform because everyone now associates AI with failure.
With strategy, AI investments compound. Shared platforms reduce cost. Shared governance reduces risk. Shared language accelerates adoption. AI strategy is the discipline of choosing — where to play, how to win, and what to say no to.
flowchart TD
V["Vision\nWhere is AI taking us?"] --> OM["Operating Model\nHow is AI organised?"]
OM --> VF["Value Framework\nHow do we measure impact?"]
VF --> MA["Maturity Assessment\nWhere are we today?"]
MA --> SP["Strategic Portfolio\nWhat do we fund first?"]
SP --> OUT["Transformation\nAI driving business outcomes"]
Try it yourself — The strategy gap
Answer these three questions about your organisation. How many gaps? That's your strategy deficit.
| Question | Your answer | What this reveals |
|---|---|---|
| What is AI for in our organisation? | If you can't answer in one sentence, strategy is missing | |
| Which AI tools are approved? | If nobody knows, governance is absent | |
| What value has AI delivered in the last 12 months? | If you can't quantify it, measurement is missing |
AI vision — where is AI taking you?
Most organisations skip the vision step. They jump straight to "which projects should we fund?" without first answering "where is AI taking us?"
An AI vision is not a mission statement. It is a concrete description of what the organisation looks like when AI is working well — in 2-3 years.
| Question | What it reveals |
|---|---|
| What will be fundamentally different about how we work? | The transformation scope |
| What capabilities will we have that we don't today? | The investment areas |
| What will our customers experience differently? | The value drivers |
| What decisions will we make differently? | The data and AI priorities |
The vision architecture has four layers: business vision (what does the business need to become?), AI vision (what role does AI play in that transformation?), data vision (what data capabilities are required?), and tech vision (what technology foundation is needed?). Each layer drives the one below it. You can't organise for what you haven't defined.
Try it yourself — Write your AI vision
Draft a one-paragraph AI vision for your organisation. If you can't fill this in, that's your first strategic gap.
| Layer | Your vision statement |
|---|---|
| Business vision — what does the business need to become? | |
| AI vision — what role does AI play? | |
| Data vision — what data capabilities are required? | |
| Tech vision — what technology foundation is needed? |
Operating model — how to organise for AI
Technology choices get the attention. But the harder, more important decision is: how is AI organised, governed, and funded in your organisation?
The operating model determines whether AI stays a side project or becomes a core capability. There are three common models:
flowchart TD
OM["AI Operating Models"]
OM --> CENT["Centralised\nSingle AI team serves the business"]
OM --> FED["Federated\nAI embedded in each business unit"]
OM --> HYB["Hybrid\nCentral platform + distributed teams"]
| Model | Pros | Cons |
|---|---|---|
| Centralised | Consistent quality, efficient resource use, strong governance | Bottleneck risk, slower response to business needs |
| Federated | Close to business context, fast iteration | Duplication, inconsistent quality, governance gaps |
| Hybrid | Platform efficiency, business agility, balanced governance | Complex to manage, requires strong coordination |
Most mature organisations converge on Hybrid:
| Component | Ownership | Why |
|---|---|---|
| AI Platform (tools, infrastructure, shared services) | Central team | Economies of scale, consistent governance |
| AI Use Cases (specific business applications) | Business units | Close to value, domain expertise |
| AI Governance (policies, risk, ethics) | Central team with business input | Consistency + context |
| AI Talent (hiring, training, career paths) | Shared model | Flexibility + retention |
The operating model answers three questions: who does what, who decides what, and who pays for what.
Try it yourself — Your operating model
For your organisation. Where are the gaps between current and should-be? That's your operating model work.
| Component | Current state | Should be |
|---|---|---|
| AI Platform — who owns tools and infrastructure? | ||
| AI Use Cases — who decides what to build? | ||
| AI Governance — who approves and oversees risk? | ||
| AI Talent — who hires, trains, develops? |
Maturity assessment — where are you today?
You can't build a roadmap without knowing where you're starting from. Most organisations overestimate their AI maturity.
A maturity assessment gives you an honest baseline across five dimensions — strategy, people, process, data, and technology — so you can prioritise the right investments.
| Level | Description | Signs |
|---|---|---|
| 1. Ad hoc | No strategy, no coordination | Individual experiments, no governance |
| 2. Aware | Leadership aware, some pilots | Strategy discussions, scattered pilots |
| 3. Defined | Strategy exists, some capabilities built | Approved tools, governance draft, training started |
| 4. Managed | Capabilities measured and governed | AI register, value tracking, systematic adoption |
| 5. Optimising | Continuous improvement, AI as competitive advantage | AI-driven products, culture shift, market differentiation |
| Dimension | Level 1 (Ad hoc) | Level 3 (Defined) | Level 5 (Optimising) |
|---|---|---|---|
| Strategy | No AI vision | Vision documented, roadmap exists | Strategy drives business transformation |
| People | No AI skills | Training programme, some specialists | AI literacy across organisation |
| Process | Manual, reactive | Some processes AI-augmented | AI-first process design |
| Data | Siloed, untrusted | Data governance in place | Real-time, integrated data fabric |
| Technology | Ad hoc tools | Approved platform | Scalable MLOps, continuous deployment |
Maturity gaps drive the roadmap — invest where you're weakest relative to where you need to be. A Level 1 organisation trying to execute a Level 5 strategy will fail. Assess honestly, then plan accordingly.
Try it yourself — Your maturity scorecard
Rate your organisation 1-5 on each dimension. Your lowest score is your biggest bottleneck. That's where to invest first.
| Dimension | Your score (1-5) | Evidence | Priority to improve |
|---|---|---|---|
| Strategy | |||
| People | |||
| Process | |||
| Data | |||
| Technology |
Value framework — how to measure AI impact
"We invested in AI but can't prove the value" is the most common complaint from AI leaders. The problem is not that AI creates no value — it's that nobody defined what value looks like before starting.
A value framework defines, before any AI initiative begins, what success looks like and how it will be measured.
flowchart TD
VF["AI Value Framework"]
VF --> FIN["Financial Value\nRevenue · Cost savings · Productivity"]
VF --> CUST["Customer Value\nExperience · Speed · Personalisation"]
VF --> RISK["Risk Value\nCompliance · Safety · Reputation"]
VF --> STRAT["Strategic Value\nCapability building · Optionality"]
| Value type | Examples | Measurement approach |
|---|---|---|
| Financial | Cost reduction, revenue increase | Direct financial metrics |
| Customer | Better experience, faster service | Customer satisfaction metrics |
| Risk | Fewer incidents, better compliance | Risk metrics, audit results |
| Strategic | New capabilities, competitive positioning | Capability assessments, market metrics |
The value chain connects AI investment to business outcomes: AI Activity → AI Output → Business Outcome → Business Value. The gap between "output generated" and "business outcome" is where most AI programmes lose the plot. Value metrics feed the strategic portfolio — fund what creates the most value. Value reporting builds executive confidence — which sustains investment.
Strategic portfolio — what to fund first
Every team has AI ideas. Resources are finite. Without a portfolio approach, the loudest voice wins — not the highest value.
Strategic portfolio management applies investment logic to AI initiatives — balancing value, risk, feasibility, and strategic alignment.
flowchart TD
IDEAS["AI Ideas\nfrom across the organisation"] --> SCREEN["Strategic Screen\nDoes it align with the vision?"]
SCREEN --> ASSESS["Value Assessment\nWhat's the expected return?"]
ASSESS --> PRIORITISE["Prioritise\nBalance quick wins + strategic bets"]
PRIORITISE --> FUND["Fund\nApproved initiatives get resources"]
FUND --> TRACK["Track\nMeasure actual vs. expected value"]
TRACK --> LEARN["Learn\nFeed insights back into portfolio"]
The portfolio should balance three categories:
| Category | Purpose | Example |
|---|---|---|
| Quick wins | Build credibility, show value fast | AI-assisted email drafting, meeting summaries |
| Core improvements | Improve existing processes | Predictive maintenance, demand forecasting |
| Strategic bets | Create new capabilities or markets | AI-powered products, autonomous systems |
Prioritisation criteria: strategic alignment (25%), value potential (30%), feasibility (20%), risk (15%), learning value (10%). Portfolio decisions are where strategy meets execution. Maturity assessment determines feasibility scores. Value framework determines value scores. Operating model determines who makes portfolio decisions.
Try it yourself — Your portfolio
List every AI initiative your organisation is running or planning. Score each against the five criteria. Which ones score highest? Those are your priorities. Which ones score lowest? Those are your candidates for stopping.
| Initiative | Strategic alignment (25%) | Value potential (30%) | Feasibility (20%) | Risk (15%) | Learning value (10%) | Total |
|---|---|---|---|---|---|---|
Building the roadmap
Strategy without a roadmap is a poster on the wall. The roadmap translates strategy into a sequenced plan of action.
An effective AI roadmap sequences investments across three horizons:
| Horizon | Focus | Key activities |
|---|---|---|
| H1: Foundation (0-6 months) | Build the base | Governance framework, approved tools, data readiness, quick wins |
| H2: Scale (6-18 months) | Prove and expand | Core use cases, platform build-out, talent development |
| H3: Transform (18-36 months) | Strategic advantage | New AI-powered products, business model innovation |
The roadmap is the output of vision + operating model + maturity + portfolio. Roadmap phases align with governance maturity — ad hoc to managed to optimising. Roadmap review cadence keeps strategy alive — quarterly reviews, annual refresh.
Try it yourself — Your three-horizon roadmap
Map your initiatives across three horizons. Do you have initiatives in all three horizons? If everything is in H1, you're not thinking strategically. If everything is in H3, you're not delivering value now.
| Horizon | Timeframe | Your initiatives | Success criteria |
|---|---|---|---|
| H1: Foundation | 0-6 months | ||
| H2: Scale | 6-18 months | ||
| H3: Transform | 18-36 months |
The strategy stack in one diagram
flowchart TD
V["Vision\nWhere are we going?"] --> OM["Operating Model\nHow do we organise?"]
OM --> MA["Maturity\nWhere are we today?"]
MA --> VF["Value Framework\nHow do we measure?"]
VF --> SP["Portfolio\nWhat do we fund?"]
SP --> ROAD["Roadmap\nIn what order?"]
ROAD --> EXEC["Execution\nAI driving business outcomes"]
The strategy:
Strategy before technology. The hardest decisions are not which model to use — they're where to invest, how to organise, and what to say no to. Without that clarity, every team experiments in isolation and nobody learns from each other.
Cheat sheet — all the key terms
| Term | Plain English | Where it fits |
|---|---|---|
| AI Strategy | Choosing where to bet and building the capability to win | Every AI programme |
| AI Vision | A clear picture of the future state, not a technology list | Foundation |
| Operating Model | Who does what, who decides, who pays | Organisation design |
| Centralised | Single AI team serves the business | Operating model choice |
| Federated | AI embedded in each business unit | Operating model choice |
| Hybrid | Central platform + distributed teams | Operating model choice |
| Maturity Assessment | Honest diagnosis before prescription | Baseline |
| Value Framework | Define the win before you start playing | Measurement |
| Portfolio Management | Investment logic applied to AI ideas | Prioritisation |
| Quick wins | Build credibility, show value fast | Portfolio category |
| Strategic bets | Create new capabilities or markets | Portfolio category |
| Roadmap | Sequenced bets across three horizons | Execution plan |
How to know if this landed
You'll know this has landed when someone can articulate the AI vision in one paragraph — and every team member can too. When the operating model is documented with clear roles, responsibilities, and funding model. When a maturity assessment has been completed across all five dimensions. When a value framework is defined with at least three measurable outcomes per initiative. When a portfolio is managed with explicit prioritisation criteria — not loudest voice. When a roadmap is reviewed quarterly with progress against milestones. When executive sponsorship is active, not ceremonial. And when at least one AI initiative is delivering measurable business value within 6 months.
What leadership teams realise in the room
The portfolio exercise is the one that shifts behaviour most immediately.
I ask leadership teams to bring their real AI initiatives — every pilot, every proof-of-concept, every idea on the backlog. The number is consistently higher than anyone expected. I've worked with organisations that had 15+ disconnected pilots, no shared infrastructure, and no one who could say what the total investment was.
The screening conversation changes everything. "Does this align with the vision?" "What's the expected return?" "Do we have the data, skills, and technology?" "What could go wrong?" "What do we learn even if it doesn't fully succeed?" These questions, applied consistently, reduce the portfolio from 15 pilots to 6 prioritised initiatives with clear owners.
The vision alignment exercise tends to resolve a conversation that's been happening for months: what are we actually trying to do with AI? After walking through the four vision layers — business, AI, data, technology — teams usually produce a one-paragraph vision statement that everyone can agree on and repeat. That's the moment the strategy becomes real.
The maturity assessment produces the most uncomfortable moments. Teams that believed they were Level 3 discover they're Level 2. The gap between "we have a strategy" and "we have a strategy, approved tools, governance draft, and training started" is usually wider than anyone expected. That's not a failure — it's clarity. And clarity is what lets you plan honestly.
The roadmap conversation tends to surface tensions that were already there but unnamed. Teams that have been trying to do everything at once suddenly see the logic of sequencing: foundation first, then scale, then transform. The three-horizon model gives them permission to do quick wins now while planning strategic bets for later. That's not bureaucracy. That's the infrastructure for moving faster with less risk.
Book a Workshop
Ready to build the AI strategy your organisation needs — not just more pilots?
or
2-day workshop + coaching includes vision alignment and operating model design, maturity assessment with your leadership team, value framework tailored to your industry, portfolio assessment of your current and planned AI initiatives, 12-month roadmap with clear milestones and owners, and governance starter pack for immediate implementation.