Forty-one systems. Twelve we can talk about.
Every entry names the mechanism, the measurement, and what we would do differently. The rest are under NDA — ask and we will walk you through them on a call.
Atlas Freight — dispatch that explains itself
A routing assistant that proposes, shows its reasoning trace, and waits for a dispatcher. Eleven months in production across four depots.
Read the case studyVero Health — triage notes in 40 seconds
Clinician-reviewed summarisation with a full audit trail on every generated line.
Read the case studyNorthbank — reconciliation without the spreadsheet
Nightly matching across seven ledgers, with exceptions routed to a named analyst.
Read the case studyMuni Grid — outage triage for a two-person desk
Sensor fusion and a ranked work queue that a duty operator can override in one click.
Read the case studyFenwick & Roe — disclosure review with citations
Retrieval over 2.1M documents where every answer links back to the source page.
Read the case studyKessler Rail — maintenance windows that hold
Forecasting plus a scheduling UI that makes the constraint visible before it bites.
Read the case studyHalden Labs — an eval harness they now run themselves
The measurement layer first: datasets, scoring, and a weekly regression report.
Read the case studyWant the version of this with your data in it?
We will show you the two projects closest to your problem, including the parts that went badly.