Research
Research that ships into our platforms
Agents, scientific models, funding intelligence and AI governance.
Research areas
Agent orchestration
Runtimes that coordinate many agents reliably, with low memory and compute overhead and graceful handling of failure.
Scientific language models
Compact, calibrated models for reading and screening research literature on everyday hardware.
Research and funding intelligence
Graphs that connect funders, programmes, organisations and outcomes across countries.
AI governance
Registers, risk evidence and audit methods that make AI accountable in public institutions.
Publications and releases
Ark-1
Our agent-orchestration model.
Numo-1-E2B
Open model for scientific literature screening. DOI 10.57967/hf/11188.
The full AI stack: India, Europe, and the race to build a third force
A seven-layer map of the AI stack and its chokepoints.
Agentpower
A unit of cognitive work output for the AI age, released into the public domain.
Collaborate with us
We work with universities, laboratories and consortia on joint research and European programmes, with compute from European supercomputing centres.