§ Case study · AI strategy
An AI opportunity portfolio for a European insurer.
A prioritised opportunity portfolio and an operating model to run it — grounded in what the business could actually build in the next three quarters.
a European insurer
Client
Enterprise
Tier
6 weeks
Duration
2026
Delivered
The context
What they came to us with.
Twelve business units, each running its own AI pilot, none of them talking to each other. Group leadership had asked for an AI strategy; the head of technology had asked for something that wouldn't add three more platforms to the platform team's roadmap.
Their previous strategy engagement — a big-consultancy deck a year earlier — had recommended twenty-two "priority" use cases, each described in a paragraph, none of them costed against what the business could realistically build. Half were unbuildable at the claimed cost; the other half had already been shipped by a business unit and the deck hadn't noticed.
They wanted a portfolio that would survive contact with their own engineering leadership and a written operating model the board could approve without a translator.
The approach
What we did, in order.
-
One-page opportunity write-ups.
For every candidate use case, a one-page document that answered three questions: what would this look like if we built it well; what does "good" look like at Year 2; and who else has done something shaped like this. We wrote 23 of these.
-
Practitioner scoring.
Each opportunity was scored by three signals — buildable inside a realistic quarter, defensible against next year's model economics, and adjacent to a capability the group actually had. Scoring was done by us alongside two of their senior engineers.
-
Sequencing for compounding.
The final six were sequenced not by NPV but by what capability each would leave behind for the next. The first two built the data and platform muscle the next four would need.
-
Operating model on one page.
Governance, funding, ownership, capability sourcing, vendor relations, risk. Not organograms — a written description of who decides what, with what evidence, at what pace.
-
A board narrative anyone could read out loud.
One document. One-slide summary. Practitioner-tested language. The version leadership could deliver in a room without a translator or a decoder ring.
How we measured
The bar we held ourselves to.
- Every recommended use case was reviewed by an engineer who would be plausibly staffed on it.
- Every operating-model recommendation was pressure-tested against a concrete decision the client had made in the previous six months — did it make that decision cleaner?
- The final narrative was read aloud to a mock audience of three unaffiliated senior stakeholders; every ambiguity they raised was fixed before the board version.
The outcome
Where it landed.
23
Opportunities surveyed
6
Prioritised for Year 1
1
Board-approved operating model
The board approved the operating model at the end of the engagement and the first sequenced use case entered discovery two weeks later. Two other use cases from the portfolio were re-shaped in light of the sequence — one was pulled forward, one was pushed back a quarter to let a platform investment land first.
Twelve months on, four of the six sequenced use cases have shipped and the platform investments made in the first two are being reused by three of the twelve business units that weren't originally involved in the portfolio.
Retrospective
What we'd do differently.
- We started with the portfolio and back-filled the operating model. We'd flip it — the operating model shapes which opportunities are even worth surveying.
- We would spend a day earlier with finance. The costing framework we used was engineer-native; finance had to translate it and lost some of the argument in the translation.
- The one-page write-ups were the most valuable artefact. We would produce them as a permanent template the client could use for future opportunities, not as an engagement one-off.
Services used
The moves in this engagement.
S5 · Advise
Technical & Solutions Architecture
A senior architect embedded with your programme — end-to-end design, build-vs-buy calls, vendor evaluation, technical risk review.
S6 · Advise
Strategic AI Consultancy
Leadership-level direction — opportunity portfolio, operating model, board-ready narrative — grounded in what's actually buildable today.
Related work
Other engagements in the same shape.
§ Case study · Agentic systems
An agentic ops-triage copilot inside a UK retail bank.
Analyst-in-the-loop triage across three legacy ticketing systems — mean time to first response cut by more than half, without a single new ticketing surface.
§ Case study · LLM integration
Editorial-assist RAG for a global media company.
A retrieval-assisted drafting surface for a 300-strong editorial team — answers cited to the archive, hallucinations effectively eliminated on tested queries.
Have something with this shape?
A 30-minute discovery call. We'll say honestly whether the pattern fits.