Our GenAI Strategy Playbook helps leadership teams design their business for GenAI, rather than bolting AI on to what they already do. In short: start from the outcomes your business must produce, and work backwards.
Get the full playbook, GenAI Strategy Playbook
Start from outcomes, not processes
Most GenAI initiatives start by auditing existing processes and bolting on AI where it fits. That delivers quick wins, but it locks you into processes designed for a pre-AI world. You end up doing the same thing, just slightly faster.
The playbook opens with a thought experiment. Your best people leave tomorrow, team up with a GenAI specialist and build a competitor from scratch, with no legacy systems, no existing teams and no cultural baggage. What business do they create? What data does it need? Where do humans stay in the loop, and where don’t they? That’s the business you should be building.
Five layers, one outcome at a time
The framework works through five layers in sequence, each building on the last: functional outcomes, principal data points, human-in-the-loop, guardrails, and defensibility and IP. You pick one functional outcome and take it through all five before starting the next.
Each layer comes with practical tests. The format test separates a real outcome from a process output dressed up as one. The new entrant test strips your data requirements back to what you’d need if you were starting tomorrow. And four factors – regulation, liability, cost of error and relationship value – decide where a human must stay.
Defensibility runs through every layer. GenAI turns capability into a commodity, so what’s left to defend is the institutional knowledge a competitor can’t access and would take years to build.
The with/without test, the hiring test and the investor test help you tell genuine IP from institutional habit, and structural moats from timing advantages that erode within 18 months.
In the full playbook, you’ll find step-by-step guidance and a checklist for each layer, worked examples from advice, claims and mortgage lending, a defensibility profile to close the loop, and a method for deciding which outcome to deploy first and which to tackle next.
