semantic state
How a system represents what it believes and how that steers the next decision.
an ai research lab · est. 2026
thought & memory,working together
- mission
- Decoding human cognition across environments to engineer controllable ai decision systems.
- vision
- A future where intelligent systems make decisions from explicit, updatable state.
01 clusterwords · a daily puzzle
can a word game show how you think?
Clusterwords is our daily connections-style puzzle and our microscope for how a mind, human or machine, moves from uncertainty to a committed decision.
02 the lab
two ravens, one mind
We build systems that think and remember. Huginn reasons, Muninn recalls. The future of intelligence is the conversation between them.
huginn · thought
Reasons over the present: representation and state that make decisions reliable and controllable.
muninn · memory
Recalls across time: internal state that keeps decisions consistent from one moment to the next.
AI systems generate outputs from implicit context, but reliable decisions require explicit state. We study how cognition holds up across environments, and design the representations and memory that make decisions consistent and controllable over time.
03 what we're working on
the work
memory
What gets carried forward, what fades, and how recall stays faithful to what happened.
controllable decisions
Behavior that's consistent and steerable over time, not re-rolled from implicit context each turn.
04 first research result
gentags
05 support
back the lab
We're independent and small. If the work or the games resonate, you can help keep the lights on one-time or recurring.
support the lab · soon
Stripe-hosted · no card data touches this site.
06 contact
say hello
Or just email hello@hugmun.ai.