Research
We publish the results of our applied research. Where work for client companies produces a generalizable artifact — a method, a framework, an algorithm — we write it up as a paper and release it publicly.
Research streams
Distributed AI
Algorithms that run from the industrial server to the edge device. Vector databases for semantic memory. Federated learning without centralizing data.
Agentic systems
Multi-model architectures with traceable policy, permissions, and audit. Visual orchestration (Reasonance) and codified discipline (VIBE Framework).
Sovereign on-premise AI
Deployment patterns for data that cannot leave the organization. Hybrid where it makes sense, on-prem where it must. Fine-tuning on proprietary data when needed.
AI engineering on legacy systems
Auditing, architectural refoundation, stabilization of AI codebases that grew too fast. Senior engineering applied to what already exists.