Frontier research, in the open.
The AGI Scientist is an open lab for the road to AGI — reproducible experiments, interoperable tools, and the people turning questions into work that ships.
- 2
- Researchers & builders
- 10
- Open publications
- 6
- Tools in the registry
- 100%
- Open by default
Start with an open question
Every breakthrough begins in the open. Post a hypothesis, frame a problem, and invite the community to sharpen it with you.
Run reproducible experiments
Pin your environment, data, and code. Results others can re-run are results others can build on.
Turn findings into tools
Graduate experiments into interoperable tools and projects that anyone can plug into their own stack.
Publish, review, repeat
Ship it in the open, get peer-reviewed, earn credit — and hand the next researcher a higher starting point.
Everything the loop needs, in one place.
From the first question to a shipped tool — open, interoperable, and built with the community, not behind a login wall.
Research
Reproducible experiments and living papers, published in the open and built to be challenged — not press-released.
Open Projects
Ambitious projects built in public. Fork, contribute, and ship, with credit and governance in the open from day one.
Tools
A registry of agents, models, and utilities — versioned, documented, and ready to plug into your own stack.
Interoperability
Open standards and adapters so tools, datasets, and agents speak the same language across the whole stack.
Community
Working groups, peer review, and collaborators. Bring a question; leave with people to answer it alongside.
Knowledge
Guides, primers, and curated reading paths — a clear route from first principles to the frontier.
Open experiments, living papers
Reproducible, peer-reviewed, and free to read. Publish your findings and let the community build on them.
InterpretabilityMapping latent goals in open-weight agents
A reproducible probe of emergent planning circuits across three model families.
11 min read
BenchmarksBeyond accuracy: measuring scientific reasoning
Why current evals miss the point, and an open protocol that doesn't.
8 min read
AlignmentCooperative protocols for multi-agent labs
Designing incentives so autonomous researchers share, not hoard.
14 min readAmbitious projects, built in public
Fork, contribute, and ship. Governance and credit are built in, so collaboration stays fair.
OpenLab
A self-hosting autonomous research environment that runs and documents its own experiments.
View projectInterop Registry
Shared schema + adapters letting any agent discover and call any compliant tool.
View projectReproduce
One-click reproduction of published results, pinned to exact environments and data.
View projectA registry where everything connects
Vetted, versioned tools that speak a common protocol. Discover, plug in, and compose.
AgentMesh
OrchestrationCompose multi-agent pipelines visually.
EvalKit
EvaluationReusable, sharable benchmark harness.
DataLoom
DataVersion, clean, and trace datasets.
ProbeLens
InterpretabilityInspect activations in any model.
GovDoc
GovernancePolicy-as-code for autonomous systems.
SafeRun
SafetySandboxed execution for agent code.
Why openThe most important technology of our time shouldn't be built behind closed doors.
Reproducible results, shared tools, and credit in the open aren't idealism — they're how the field moves faster and stays accountable. That's the whole bet.
Bring a question. Leave with people.
Scientists, engineers, and builders — across labs, startups, and bedrooms — working in the open. Join a working group, share what you're stuck on, and find collaborators for the next step.
Working groups
Self-organizing teams around interpretability, evals, agents, and safety.
Mentorship & peer review
Get feedback from people who've shipped — and give it back.
Idea to funded project
Pitch in the open, find collaborators, and graduate into incubation.
Everything is open. Support the mission.
Open research stays free, forever — no tiers, no paywall. If you want to help keep it that way, here's how.
Read & share
Read the open research, put it to work, and pass it on. Reach is what keeps open science worth doing.
Star us on GitHub
Our projects and tools are open source. A star helps others find the work — and pull requests help even more.
Get the briefing
Join the weekly newsletter. No spam, no paywall — just the open path to AGI, every week.
Sponsor the lab
Funding the compute, the events, and the people. Sponsorships open soon — say hello if you'd like to help.
One email a week. Only signal.
Open research, new tools, and the people building the road to AGI — distilled into one short read. No spam, unsubscribe anytime.
Free to join · No spam · Unsubscribe anytime