
Artificial intelligence
Decisions backed by data
The method we follow on AI projects: identify where the impact is, control the data sources, and keep the final decision with the person accountable for it
The challenge
Data accumulates inside organisations without reaching the people who decide. Rushed pilots either overreach on privacy, or end in a recommendation nobody can trace back to its source
Our approach
We start from the question rather than the tool: we narrow the data sources, make every output traceable and reviewable, and let no consequential action run without human approval
Results
Defined use cases, each with an owner and a measure of success
The least sensitive data possible, in the pilot and in production alike
Every output traceable to its source, and every recommendation carrying its reasoning