Work that made it to production.
The problem we were handed, what we built, and what changed. Clients are anonymised; the engineering is not.
- 01Security operations
An agentic security operations centre that clears the queue and shows its working.
An enterprise security operations team was drowning in alerts. We built a verifier-graded agent system that triages, enriches, and closes the routine ones, so analysts spend their shift on real incidents.
Read the case study72%Of routine alerts triaged and closed without an analyst11 minMedian time to acknowledge, down from several hours - 02Banking & finance
A small language model that profiles every business borrower and watches for what changes.
A bank's credit risk team was assembling business risk profiles by hand from financials, ownership records, internal exposure, and external signals, then not looking again until the next annual review. We trained a small language model that reads those sources inside the bank and keeps an evidence-backed profile and monitoring alerts on every borrower.
Read the case study70%Less analyst time per borrower review24 hrsFrom a new signal appearing to a monitoring alert, instead of the next annual review - 03Industrial safety
VR safety training that lets workers rehearse the hazard before they meet it.
An industrial operator needed to certify workers on high-hazard procedures faster, across many sites, without putting anyone at risk to practise. We built VR training on a digital twin of their own plant, where every step is scored against the written procedure.
Read the case study60%Less time to certify a new worker on a high-hazard procedure9 of 10Trainees pass the live assessment first time, up from roughly half