Technology · Financial Services · North America
Sequencing an AI programme behind data readiness
A board-approved AI mandate with no agreed link between the models proposed and the decisions they were meant to improve. The programme was resequenced around data readiness before further spend.
North American financial services firm, regulated environment, board-level AI mandate, three pilots already in flight.
- Pilots rationalised
- 3 → 1
- Programme cost avoided
- ~40%
- Engagement length
- 9 weeks
The situation
Budget had been approved on the strength of the category. Delivery teams inherited an ambition without a foundation.
Three pilots were already running, none of them attached to a decision the business could act on.
Our approach
We started from the decisions the business wanted to improve and worked backwards to the data those decisions require.
Each pilot was assessed against data availability, quality, and ownership before any tooling question was reopened.
The outcome
Two pilots were stopped and one was rescoped around a decision with measurable value.
A staged data remediation plan replaced the original tooling roadmap, at materially lower cost than discovering the gap mid-programme.
The client is not named and details are anonymised. Figures are representative of the type of business consultation First Osman performs.