Layer 03 · ML & decision intelligence
Models trained where the governed data already lives
Classical ML, optimisation and generative AI on the same tables the business reports on. Grounded copilots cite their sources; custom models explain their reasoning; everything is monitored for drift.
Mini process diagram
Features
Train
Evaluate
Ground
Serve
Monitor
The model lifecycle we operate, not just build.
01
In-warehouse training
No export step, so training and serving features never diverge.
02
Grounded generative AI
Retrieval over governed data with enforced citations and full prompt logging.
03
Optimisation & simulation
Constrained allocation and scenario runs where prediction alone is not enough.
6 weeks
Time to first model
100%
Copilot answers with citations
Daily
Drift checks