Research & engineering
An AI lab in Delft.
The lab runs itself.
Agents research, build, test, and coordinate. Vlad Iftode leads the direction.
Ask about the work, or bring us a problem.
Founded and led by Vlad Iftode.
See what we’re working onWork from the lab
Explore all the workThe lab at work
Agents, working together.
A question becomes research, an implementation, and a review. Specialist agents pass the work along, with the context the next one needs.
MCP integrations connect the toolsA visual study of shared work and context.
Identity and trust
TrustChain
Signed interactions make relationships between agents verifiable. Identity, delegation, and trust, built in Rust and described in a public Internet-Draft.
A visual study of two sides of a signed relationship.
Models and experiments
Understanding what works.
Diffusion sampling, federated learning, and recommendation systems. We reproduce methods, investigate their behavior, and test what makes a difference.
A visual study of iteration and convergence.
Founder
Founded and led by Vlad Iftode, with AI studies and research at TU Delft. He sets the direction of the lab and stays close to the work.
Client work stays private.
Direction
Autonomous agents, shared context
Our agents keep context, use tools, and pass work between each other. Research, engineering, and verification are connected inside the same lab.
Each project adds a piece: a better connector, a model we understand, a way to know which agent acted. This is where those pieces come together.
How the lab worksHave something to build?
Tell us what you want to make. A rough idea, a stubborn problem, or something you want to try.