1st Place
polar bears
TRY THE DEMO LIVE!
Core trick
Builds a MuJoCo bench twin from a photo, simulates the protocol stepwise, and traces each failed well to its causal step.
Description
TRY THE DEMO LIVE! labtwin.vercel.app
note: the hosted demo replays a cached run. non-demo data requires requires a local server and API key: please see the README or contact us!
LabTwin is a digital twin world model for the wet lab: Gemini 4 Argon and a swarm of Managed Agents turn a biolab protocol and one photo of your bench into a physics-based 3D digital twin, run every step before a real pipette moves, and fix the mistake before anyone spends a single sample.
You give it three things: a protocol (PDF or Markdown), a photo or short video of your lab's tools/layout, and the result you want in each well. Gemini 4 Argon reads the protocol and the photo together and writes a spec of your bench. An orchestrator then fans out Antigravity Managed Agents in parallel sandboxes, and they write a physics twin of your bench: a whole-room MuJoCo scene, the liquid state, the reward and a Gymnasium RL environment. Argon renders the twin from where your photo was taken, critiques it against the photo and fixes itself; in our demo the visual match went from 54 to 88.
The twin then runs your protocol step by step. It compares every well’s volume and concentration with your target and traces each failed well back to the exact step that caused it. Say “Argon, fix it”: Argon rewrites the protocol with a reason for each change, and the simulator re-runs it from scratch to prove the fix.
Every twin is also a real RL environment: an agent trained from scratch reaches a 1.0 success rate. You can also play the protocol yourself in 3D, with mistakes flagged as you click.
README excerpt
# LabTwin > README drafted by Gemini 4 Argon from the team's contract and notes, then fact-checked against the repo by Agent B. **LabTwin is a physics-verified biolab protocol debugger and digital-twin generator powered by Gemini 4 Argon.** A scientist gives LabTwin a liquid-handling protocol (PDF or Markdown), a photo or short video of their lab bench, and the target result they expect in each well. Gemini 4 Argon (`gemini-4-argon` via Google's Interactions API) reads the protocol and bench photo, builds a 3D MuJoCo digital twin and Gymnasium environment of that specific bench, and runs the protocol in simulation to show the exact step and well where it fails to produce the target volumes or concentrations. Argon then rewrites the protocol and re-runs it in the physics twin to prove the fix works—or the scientist can click through the protocol themselves in the 3D twin and see mistakes flagged live. ## Who It Is For - **Wet-lab scientists** who want to check whether a written pipetting procedure (such as an 8-well 1:10 serial dilution across row A of a 96-well plate) actually produces their target well volumes and concentrations before touching real samples and reagents. - **Lab-automation engineers** who need a validated MuJoCo physics twin and Gymnasium RL environment for a new protocol on a specific bench before running it on a real liquid-handling robot—work that normally takes weeks of hand-written MuJoCo XML and reward code. ## The 3-Minute Demo Flow 1. **Upload