The Proto team from @BrianHie’s lab will be at the re:AGENT hackathon, a two-day weekend (8/15–8/16) in San Francisco for people interested in building AI scientists, datasets, or pipelines for biological design.

Apply by Thursday, July 30th, 12 PM PT at https://t.co/H2ddXzQBLP https://t.co/rWcFQ5uF3X

re:AGENT – End to End Agentic Science · Luma

Source: https://luma.com/g6org075 ​WHAT IS re:AGENT

re:AGENT is a two-day build weekend focused on the infrastructure scientific agents still need: better datasets, sharper tools, and reliable ways to evaluate their work. Participants will usePaperclip,Proto,Claude, and other scientific AI tools to build systems that can take on real scientific problems and produce results worth trusting.

​Paperclip turns papers, datasets, clinical trials, and biological databases into an agent-native workspace that AI systems can search, synthesize, and build upon. Proto is a design framework and infrastructure layer for generative biology, allowing builders to combine AI models and search for DNA, RNA, or protein sequences that satisfy custom criteria.

​The weekend brings together engineers, scientists, computational biologists, founders, and AI researchers. Projects might include assembling a research-ready dataset that would be impractical to create by hand, along with the analysis needed to show that its findings are real. Others might involve designing a new biological system to a defined specification and testing whether it holds up against existing evidence.

​The aim is not just to accelerate existing workflows, but to create the tools, datasets, and evaluation methods that make entirely new scientific questions and experiments possible.

​READY TO BUILD?

​If you’re excited about the future of AI, agents and scientific discovery, we’ll see you there...

​We will close applicationson Thursday, 30th of July, 12 PM PST.Spots will be confirmedon Friday, 31st of July.

​WHO SHOULD ATTEND

​**People with the questions:**scientists of all stripes. Wet lab, computational, clinical, industry. Biology, biochemistry, biophysics, chemistry, structural. Anyone with a real problem, ideally a real dataset, and the judgment to tell right from wrong results.

​**People who build for a living:**ML engineers, software developers, and agent builders who can take a half-formed scientific problem and ship something that runs by Sunday.

​

No prior AI expertise required. Participants are welcome at all levels.

SCHEDULE

​Day 1 — Saturday, August 15 (from 8:30 am)

• ​Check-in, coffee & bagels
• ​Welcome, tracks & host tool lightning talks
• ​Ideation, pitches & team formation
• ​Build sessions, with mentor office hours throughout
• ​Dinner & sponsor remarks
• ​Progress checkpoint, then Day 1 close

​Day 2 — Sunday, August 16 (from 8:30 am)

• ​Breakfast
• ​Final build session
• ​Submissions due, brunch & project expo
• ​Featured talks on the future of scientific agents
• ​Project demos & live judging
• ​Awards, group photo & closing

​Tracks

​Track A: Build an AI Scientist

​Build an AI agent that can carry out a defined scientific or drug-development workflow from start to finish. It should gather evidence, use relevant tools or databases, generate and test hypotheses, produce a structured output, and make its reasoning easy to inspect. This could mean reasoning over experimental data, scientific literature, or learned representations like protein embeddings. Examples include a virtual FDA reviewer, toxicology agent, clinical trial designer, experimental planning agent, or a protein-discovery agent that generates and tests hypotheses over protein embeddings and structural features.

​Track B: Build a Dataset or Meta-Analysis

What have you always wanted the literature to tell you, but could never read fast enough to find out? Draft the queries, run them across thousands of papers, sharpen and re-run until the results hold the specific pieces you care about. Then the real work: find the pattern in what you assembled that no single paper could show you, and demo that.

​Track C: Build the Biological Design

Assemble a pipeline that designs biology to a spec you choose, from a single protein up to a multi-gene system. You set the constraints; the models generate candidates and evolve them toward something that could hold up in a real cell. Could the literature reveal the most promising design space for a generative model to explore? What you demo is the new biological design you made — the sequence or system that didn’t exist before you built it.

​Or come with a different project of your own.

​Prizes will be awarded.

FAQs

​Do I need a biology background or an AI background to participate?

No. We’re looking for engineers with no life sciences experience as well as individuals who are in life sciences. One is helpful, but you don’t need both. The best teams usually pair technical builders with someone who knows the science.

​Do I need a team before applying?

​No. You can apply as a team, but you don’t have to. You can apply solo, and you’ll meet potential teammates at the kickoff event.

​What will I have access to during the hackathon?

FreePaperclipAPI access, access toProto,ClaudeAPI credits, and compute and credits from our sponsors.

​Is there a cost to attend?

​The event is free to attend. Food, drinks, and compute are covered**.**

​What happens to what I build afterward?

You keep it. Teams are free to open-source, continue building, or spin their project into something bigger.

CO-HOSTS

​Supported by an incredible group of partners helping accelerate innovation in AI and science.

• ​GXL
• ​Arc Institute
• ​Anthropic
• ​BenchFlow
• ​future.bio
• ​Biohub