Don't build harnesses, build environments.
That's the key lesson from #EinsteinArena. We created an agent-native research ecosystem—forums, verifiers, shared infra, etc—and opened it to any AI agent.
Together, the agents made major advances on many open science problems.
Users praise EinsteinArena for letting AI agents collaborate on open science problems like the Kissing problem to achieve the biggest advance in decades that no model could make alone.
Agents collaborating on EinsteinArena were able to make the biggest advance in the Kissing # problem in decades that no model was able to do alone.
Paper: https://arxiv.org/abs/2606.10402 Arena: https://einsteinarena.com/
Great job @federicobianchy @ykwon_0407 @togethercompute and @aneeshpappu
Very excited about the direction of agent collaboration for discovery. Awesome work led by @federicobianchy @ykwon_0407 and advised by @james_y_zou! I had a great time helping analyze agent collaboration patterns.
Paper: https://arxiv.org/abs/2606.10402v1
Don't build harnesses, build environments.
That's the key lesson from #EinsteinArena. We created an agent-native research ecosystem—forums, verifiers, shared infra, etc—and opened it to any AI agent.
Together, the agents made major advances on many open science problems.
@james_y_zou Are there any other examples of useful environments? It seems like everybody's building harnesses instead
Don't build harnesses, build environments.
That's the key lesson from #EinsteinArena. We created an agent-native research ecosystem—forums, verifiers, shared infra, etc—and opened it to any AI agent.
Together, the agents made major advances on many open science problems.