Screen and simulate before the bench.

Test more ideas before you pipette once.

  • Docking and virtual screening against your target.
  • ADMET and PK predictions for each candidate.
  • Each run keeps its inputs, parameters, model and version.

A ligand docks into the pocket · candidates ranked

Three agents, one result.

You ask · agents take turns · results arrive cited · you sign off

  • Reader long-context LLM

    Collects known binders, targets and assay results from the literature.

  • Analyst code + structure models

    Prepares structures, runs docking and property models, ranks the results.

  • Writer drafting LLM

    Writes the screening summary with methods and sources.

What you get back.

Ranked candidates

Scored and sorted, with the model behind each score.

Property profiles

ADMET and PK predictions side by side.

Run record

Inputs, parameters and versions for every run.

Screen summary

Methods, results and sources in one report.

Common questions.

Which models can I run?

Docking, ADMET, pathway and PK models, together with trained research models for binding affinity and variant effect. Each model shows its reported validation on its own card, so you know what it was tested on before you use it.

Can I bring my own compounds and structures?

Yes. Add them to your project and the agents use them with the curated libraries. Your data stays inside the project, is visible only to the roles you choose and never trains shared models.

Can a run be repeated later?

Yes. Each run is stored with its inputs, parameters and model version. You can run it again, change one setting and compare the two results side by side.

Bring your next question.

Tell us your field and what you want to do. We set up the workspace around it.

Request access