Fish Food 657: How AI supercharges strategy and planning
Stretch use cases for AI in strategy, is AI really taking jobs, the shadow AI economy, Nano Banana and what 1910 tells us about modern technology anxiety
This week’s provocation: Stretch use cases for AI in strategy development
I ran a session with a leadership team this week in which we were working through how AI could be a true partner throughout the strategy development and deployment process. One of the issues in discussing this subject is the sheer breadth of application. Similarly to the innovation process, there’s just so many ways in which AI can be integrated at every stage and so to get the true benefit we need to truly think about it as a thought partner throughout.
The value of AI in this context is not only in efficiency but in how it can help you get to places that you probably wouldn’t have got to on your own. It ‘supercharges’ strategic processes because it both catalyses planning but also enables entirely new possibilities. Over the past few months I’ve written about using AI project spaces to develop strategy, using AI as a thought partner to challenge assumptions and think differently, using synthetic personas and research to explore ideas and using AI for simulation and scenario planning in strategy.
I’ve set out a (far from comprehensive) list of the more obvious AI use cases above, using Stephen King’s classic strategy cycle to give it some structure, but I thought it may be fun to dwell on some of the more unusual or thought-stretching ways of using AI for each stage, some of which came up in my recent workshop.
Situational awareness: Where are we?
Alongside the obvious ways in which AI can synthesise and deepen customer, category, cultural and company understanding, I’m fascinated by the whole idea of being able to better interpret weak and emerging signals. Some examples include:
Competitive intent: Beyond more obvious variables (market share, pricing, distribution) AI can infer competitive intent by analysing hiring patterns, leadership speeches, company report language, patent filings, customer reviews.
Synthetic future consumers: There are some very useful ways of using synthetic personas to explore ideas, but I like the idea of creating future AI personas trained on demographic and cultural data which can simulate how emerging customer groups might behave.
Fusing signals across domains: We often analyse markets in isolation but AI can also be used to connect faint signals across unrelated domains (science, patents, culture, policy). This can reveal the first signs of different forces colliding (for example AI + regulation + consumer activism) before they converge into disruptive change.
Clustering anomalies: Rather than just tracking trends, AI can detect clusters of anomalies across customer behaviour, competitor actions, or market data which may indicate the first ripples of disruption.
Current state analysis: Why are we there?
I’m a fan of using AI to help with root cause analysis and iteratively using AI in a ‘5 whys’ approach to surface fundamental drivers and refine problem statements. But there are some other stretch use cases here too:
Causal simulations: We usually think about simulations in the context of future possibilities, but it can be interesting to feed a bunch of historical performance, leadership decisions, and cultural indicators into an AI and ask it to simulate ‘alternate histories’. What would have happened if a different decision had been made? This helps to uncover the real drivers of causality and can inform future scenarios.
Decision-path archaeology: similarly, AI can analyse board papers, financial reports, and outcomes, reconstructing past strategic decisions and surfacing decision biases, recurring blind spots, or patterns of over/under-reaction.
Narrative mapping: AI can process years of internal comms, reports, and leadership messaging to reveal the implicit stories that have guided behaviour. These hidden narratives can help explain inertia or misalignment.
Objective setting: Where could we be?
AI is pretty good at enabling better objective and goal definition but again, there’s also some more imaginative ways in which we can use it:
Counterfactual stretch goals: Asking AI to generate objectives that assume one core constraint is no longer a factor (‘If capital were unlimited, what would our 5-year goal be?’) can open up new possibilities which can then be scaled back more pragmatically.
Inverse benchmarking with synthetic competitors: We’re already familiar with developing synthetic personas but I’m also fascinated by the idea of using AI to generate ‘synthetic competitors’, or hypothetical firms which can be used to stress-test strategies or expose overlooked goals or possibilities.
Values-aligned objective generation: AI is of course pretty good at recommending financial or market-driven goals, but don’t forget that it can also synthesise (potentially more inspiring) objectives based on stated values, purpose and to a degree the cultural DNA of a business.
Strategy formulation: How could we get there?
Some of my favourite stretch techniques here involve constraints-driven and recombinant thinking:
Constraint-flipping: Inverting assumptions (‘what if we had to achieve this with half the resources?’) using AI can help to generate unconventional pathways or creative shortcuts.
Adjacency recombination: I love norm-switching as a way to break out your sector assumptions, and cross industry analogies using AI (like applying logistics optimisation techniques to healthcare strategy) can open up entirely new thinking.
Strategy sparring: Deploying multiple specialised AI personas (tech optimist, skeptic, regulator, innovator) to debate strategy options gives you a helpful range of divergent perspectives.
Measurement and tracking: Are we getting there?
Strategic drift detection: AI can be set up to continuously compare progress against stated goals and strategy both in terms of metrics tracking but also actions taken (meeting minutes, budgets, project outputs) to flag potential deviations early.
Emergent KPIs: As progress is made KPIs may need to change, so AI can propose new KPIs that may be better than pre-defined ones.
Future-back metrics: Rather than just starting with the KPIs that may seem right today, AI can generate different metrics by working backwards from desired long-term outcomes, which can help with the tracking of leading indicators of future success.
A lot of these stretch techniques are pretty nascent but they show how we’re only really scratching the surface of how AI will change strategy development and implementation. They are less obvious but their value comes from just that - they go beyond efficiency (where many strategists stop right now) and push AI into different territories like reframing, imagination, and discovery. Whilst these remain the parts of strategy that humans are strongest at, they are also ones that AI can now take to a whole new level.