We turn fragmented data into decisions you can defend.
For twenty years we've modeled hard domains — rule-driven configuration, enterprise operations, AML intelligence. We build the operational layer others now call an ontology: bespoke, on your infrastructure, under your control.
Five disciplines we engineer end to end.
Domain modeling & ontology
We map messy, multi-source reality into a coherent operational model — objects, relationships, actions. Not a data catalogue. The semantic layer your decisions actually run on.
Data fusion & graph analysis
Entity resolution and link analysis across siloed sources. We connect the records that don't share a key and surface the relationships that matter.
Investigation & decision platforms
Case management, investigative workflows, write-back to the systems of record. Tools analysts work in — not dashboards they look at.
Governed AI
LLMs and agents operating over your domain model, inside audit, evaluation and human-in-the-loop. AI that proposes; people that decide; every step on the record.
Built for regulated environments
AML, KYC, sanctions, GDPR. We've shipped where "roughly right" is a finding. Compliance and audit are designed in, not bolted on.
Skill is table stakes. Judgment is the job.
Anyone can generate code now. The hard part was never typing — it's deciding what should exist, holding the architecture, and being accountable when it ships. We treat AI as leverage on judgment, not a replacement for it. Our systems are auditable by design, because in our domains "the model said so" is not an answer.
The seeing-stone, built in Europe.
There's one famous platform for making sense of everything at once. It's American, cloud-only, and priced for governments. We build the same class of system on your terms — EU-hosted, sovereign by default, deployable on-prem, and bespoke to your domain instead of bent around someone else's platform.
A small number of engagements, run end to end.
Discovery
We start with the decision you can't make fast enough, and the data that should inform it. A fixed-scope engagement — days, not months — to map the domain and de-risk the rest.
Domain model
We build the operational ontology: entities, relationships, events, and the rules that govern them. The semantic layer your system reasons on — reviewed with your people, owned by you.
Build
Investigative tooling, data fusion, and governed AI over the model. Shipped in increments you can put in front of analysts, running on your infrastructure from day one.
Operate
We harden, document, and hand over — or stay on to run it. Audit, evaluation, and human-in-the-loop are designed in, not bolted on at the end.
Senior engineers on the work from first call to handover — not a sales team in front and juniors behind.