FROM DISCOVERY TO BUILD-READY REQUIREMENTSHOW WE DO ITGenAI Strategy Programmes
The AI strategy playbook, applied.
Who this is for
Leadership teams who want a GenAI strategy for a specific business area - lending, servicing, claims, operations - that starts from business outcomes rather than a use-case long-list. Especially useful if you already have a proof of concept and need to know whether it deserves to go further.
Our approach
GenAI initiatives often assume existing processes are mostly right, and bolt-on AI where it fits. That locks you into processes designed for a pre-AI world: you get a faster horse when you needed a car. We start from the other end, working through the five layers of our published GenAI Strategy Playbook:
Functional outcomes
What must this business produce? Not the process, not the format: the result.
1
Principal data points
What inputs are needed to get there?
2
Human-in-the-loop
Where must a human stay, and why: regulation, liability, cost of error, relationship value?
3
Guardrails
What must be true for this to be safe, compliant and trustworthy?
4
Defensibility
What do you know that a competitor with public data and off-the-shelf AI cannot replicate?
5
AI conversations usually stop at whether something can be done. The fifth question is the more valuable one: if everyone has the same models, advantage comes from what only you know - and this programme ends with a clear view of what that is.
What happens
Discovery and opportunity mapping.
Workshops and interviews with the people who do the work - relationship managers, underwriters, claims handlers, product leads - to capture expert knowledge and surface friction. High-level process mapping focused on information flow and decision points, a technology alignment session with your engineering leadership, and a prioritised, evidence-backed view of where to focus. If you have an existing proof of concept, we assess it critically against the framework: the gaps between prototype and build-readiness, the unintended consequences for the wider process, and a straight recommendation - develop or redesign.
Solution development.
Iterative design of the priority journeys, with working prototypes built in code rather than static design files - so real data entry, error states and interactions can be tested by your engineers and product managers. Requirements are documented in real time alongside the prototypes, with open questions flagged rather than buried. Your product people work in the room with us, so the ways of working transfer along with the outputs.
Path to production.
Prototypes refined to high fidelity, then translated into development-ready epics, user stories and requirements in your preferred format - authored by us, owned by you. The programme closes with a review of outcomes, lessons and a recommended roadmap for the quarter ahead.
What you walk away with
A strategic design for the business area, worked through all five layers.
Institutional knowledge, policy and playbooks captured in a structured, machine-readable form - your institutional judgement, made systematic.
Working prototypes your teams have already handled.
Requirements your developers can pick up on day one.
And a defensibility view: a clear statement of the advantage your institutional knowledge gives you, and how the design protects it.