One day. Real genomic questions. Build agents that do useful scientific work, then show the room what they found.
AI agents can plan multi-step work, call tools, run analyses and recover from failure. Biology is where that gets hard: results must be reproducible, evidence must be traceable, and an agent must know when it cannot support a claim.
ClawBio is open-source infrastructure for that work. Its versioned skills make an agent's methods inspectable, reproducible and open to challenge.
Boston is modelled on the latest ClawBio hackathon in Berlin, with the format improved using what we learned there: setup before arrival, one clear help route, faster team formation and more protected build time.
Everything runs through Slack
The ClawBio Slack community is where we work before, during and after the event. Register for the event to receive access for setup, introductions, team formation, help, demos and follow-up. The join link is not posted publicly.
#boston-general: introductions, announcements, timetable and resources
#boston-help: the single route for setup, access, data, tool and testing problems
#boston-teams: ideas, team formation, challenge selection and working links
#boston-demos: demos, evidence, limitations and what happens next
Registered participants will receive a short technical preflight through Slack. Complete it before arrival so that the day can be spent building.
What you will build
Choose a published challenge brief or bring a biological question that we can scope with you. Challenge directions include:
End the diagnostic odyssey
Build a cancer target you would defend
Ask whose genome a method fails on
Bring your own question
Your team will build an agent that can do useful scientific work, expose its evidence and state what it cannot claim. A GitHub contribution is welcome when it fits the project, but the day is about useful work and learning, not chasing a pull request count.
What you get
A full day focused on one real scientific problem
A t