About
Every agent framework built in the last five years assumes natural language as the communication medium — because the underlying models are trained for language. Not because it is the right choice for machines talking to machines.
Mission
We are training the alternative. Not as a research exercise, but as a foundation layer — the kind of thing that, once it exists, makes a qualitatively different class of applications possible.
The Boundary Layer is trained natively for inter-agent communication. It is auditable by architecture, not by add-on. It will be open source, because protocol infrastructure should be shared infrastructure.
Team
CTO / Co-CEO
Develops the Boundary Layer architecture. All training experiments. All model releases.
Co-CEO
Healthcare strategy, regulatory compliance (MDR, EU AI Act), market development.
CFO
Operations, finance, structure.
Management Advisor
Growth strategy, investor relations, defence domain access.
Open Roles
We are looking for hands-on engineers who have trained and evaluated large language models in production — not researchers, but builders.
Location
DensCor.ai is in pre-incorporation, with incorporation planned in Luxembourg. Models are trained on MeluXina (EuroHPC Joint Undertaking). Infrastructure runs on European cloud providers.
We are not a satellite of a US hyperscaler.
Contact