Updates
Follow the development of the AI scientist
This hub develops alongside the architecture itself. New layers, new work loops and new essays are published as the system is built — including the parts that turn out to be wrong.
- R1July 2026
The hub opens
- ShippedCategory definition and the map-versus-detective frame as an interactive comparison.
- ShippedArchitecture explorer: nine layers, forty-five capabilities, three depth levels.
- ShippedWork-loop explorer: omics data to target prioritization across ten stages.
- ShippedAI scientist maturity assessment with shareable results and no sign-up.
- ShippedEssays 1 and 2, browsable by topic.
- NoteThe capability model is versioned in public, so the framework can be argued with rather than taken on trust.
- R2Planned
More loops, deeper assessment
- NextAdditional work loops: dataset to biomarker, evidence to mechanism, safety-signal investigation.
- NextA more detailed capability assessment with saved organizational profiles.
- NextExternal expert perspectives and video walkthroughs.
- NextInteractive decision-grade artifact examples.
- NextComparison of system archetypes.
- R3Planned
Live demonstrations and public versioning
- NextLive or semi-live work-loop demonstrations.
- NextPublished evaluation results and benchmarks.
- NextCommunity-submitted system assessments.
- NextPublic versioning of the capability framework.