APPLIED AI SYSTEMS LAB / CONTROL PLANE
We build AI systems where every answer must trace to a source and every action must have an owner.
For consequential decisions where plausible is not enough.
Specialist agents work in parallel against shared evidence—not shared guesses.
LAB SNAPSHOT / AUG 2026
10 systems in operation or active R&D · 870+ automated test files · 78 versioned database migrations · 23 architecture decision records · 800+ commits since April 2026SELECTED SYSTEMS
See each system's work and maturity. Private data stays private.
Capital Observatory / Investor intelligence
An observatory for publicly disclosed investor activity. It watches what funds publish — portfolio pages, reports, regulatory filings — and turns every fact into a row that can be cited.
Inspect systemRHENIUM 75 / Counterparty intelligence
Establishes the material facts about a counterparty, shows the source behind every conclusion, and turns them into a decision a company can defend: proceed, proceed on conditions, or decline.
Inspect systemKnowledge systems
A research environment for navigating a large technical corpus through lexical, semantic and evidence-aware retrieval.
Inspect systemKnowledge systems
An event-driven runtime for AI operators working across messaging, structured memory and controlled action queues.
Inspect systemHOW THE SYSTEM HOLDS TOGETHER
Sources, saved state, retrieval and approval keep answers traceable and unsafe actions blocked.
Explore the architectureDIGITAL TEAM / SIX MANDATES
Six software roles share records and challenge one another.
Coordination & Operations
Research Strategy
Product & Systems Architecture
Knowledge & Data Architecture
Reliability & Security
Strategic Challenge
THE ORIGIN
This is where research met software — close to real work.
PRIVATE INTAKE / FIRST-PARTY CHANNEL
Describe the decision, available information and actions the system must never take alone.