Ruliad, an independent research lab exploring the AI tech tree
Exploring the AI tech tree, one branch at a time.
An independent research lab looking for what comes after today's models. Our own AI research engine lets us explore far more of the tree.
Slide 1 of 1: Ruliad
We think the next leap in AI will come from new ideas, not just more scale.
Most of the field is optimising the paradigm we already have. We're looking for the next one: new architectures, new ways for models to learn, and new kinds of systems built on them.
We're independent by design. We move fast, follow ideas wherever they lead, and back every claim with the work behind it.
More about the lab →Latest research
Papers, replications and technical notes.
Use what we've built
When research becomes useful, we ship it, and we release open weights where we can.
What we're working on
The open problems we think the next paradigm hinges on. The list will change as we learn.
World models
Learning how environments behave, so agents can predict and act in them.
Continual learning
Models that keep learning after training, without starting over.
Memory and continuity
Agents that remember across long runs and stay coherent over time.
Perception
Models that decide where to look and what to keep.
Reasoning
Fast and slow thinking in small models.
The next branch
Whatever else the next paradigm turns out to need. We'll add branches as we find them.
From the lab
Collaborate
Working on something related?
We take on a few collaborations where our research meets a real problem.
Get in touch →Careers
No open roles right now.
If the work resonates, write to us anyway. We read everything.
Say hello →Newsletter