Flagship case study
the platform.
Adoption-analytics platform for a flagship insurance risk product
Problem
Reporting for a flagship insurance risk portfolio ran on manual decks and hand-scraped spreadsheets: answering one question meant a fire drill across files.
Solution
A live analytics platform, taken from first working version to production, that senior leadership now runs its portfolio reviews on.
- Single sign-on, versioned migration-safe schemas and Dockerised CI/CD on AWS.
- Releases every fortnight, with priorities set by the business.
- Used weekly by senior leadership; made available to thousands across the division.
- One owner from the first version to the live platform.
Reporting had outgrown the spreadsheet.
A flagship insurance risk portfolio was being run out of manual decks and hand-scraped spreadsheets. Answering a single question meant a fire drill across files, and senior leadership had no reliable, real-time view of where things stood.
The organisation did not need another report. It needed a system leaders could trust and use in the room.
First, a working version.
The first build came directly from the underlying data: a model, an interface, and the decision to make the data something you query rather than something you assemble by hand.
That version is what the platform grew from.
Two ways to answer the same question
Assemble by hand
- Find the files
- Scrape the numbers
- Reconcile the versions
- Assemble the view
An answer, eventually. Then the question changes.
Query the platform
- Ask the question
A live answer, the same for everyone.
Then came production.
The move to production was the full treatment on AWS: single sign-on, versioned migration-safe schemas, and Dockerised delivery through a CI/CD pipeline across development, staging and production.
It stopped being a prototype and became infrastructure people depend on.
Run like a product.
Releases went out on a fortnightly cadence, with a business-led prioritisation process so the people closest to the work decided what shipped next.
That rhythm is what kept the platform trusted as it scaled.
The operating rhythm
- A release ships
- The business decides what ships next
Leadership runs on it weekly.
It is used weekly by senior leadership to run portfolio reviews, replacing hundreds of slides of manual reporting.
It was made available to thousands of colleagues across the division, and requested beyond the team it started in.
How far it travelled
Requested beyond the team it started in
Made available to thousands across the division
Senior leadership, weekly portfolio reviews
Origin, build, releases, adoption, team.
The product, the features, the release management, the operating rhythm and the line on scope stayed with one owner from origin to adoption.
The team grew with it: graduate engineers, hired, led and developed on the programme, now ship production features on it.
Selected technical detail
- AWS deployment with single sign-on and role-based access.
- Versioned, migration-safe schemas, so the data model evolves without breaking what is live.
- Dockerised CI/CD across dev, staging and production.
- A business-led prioritisation process feeding the fortnightly release cadence.
- LLM-assisted analysis on managed cloud infrastructure, with human-in-the-loop review.
Where judgement sits
The system drafts
LLM-assisted analysis over the source data
A person reviews
judgement, before anything ships
The platform shows
only what survived review
What review corrects feeds the next pass.
Used weekly by senior leadership to run portfolio reviews, replacing hundreds of slides of manual reporting.
Building AI capability inside financial services?