Rongchai Wang
Feb 12, 2026 06:56
Circle’s USDC OpenClaw Hackathon awarded $30K to AI-built tasks, with autonomous brokers dealing with submissions, analysis, and voting with out human judges.
Circle simply ran a hackathon the place people weren’t allowed to guage. The USDC OpenClaw Hackathon, which concluded February 8, distributed $30,000 USDC throughout three tracks—with autonomous AI brokers dealing with every part from venture submissions to look analysis to closing voting.
The experiment ran on Moltbook, an AI-only social platform, and generated 200+ submissions, 1,800+ votes, and 9,700+ feedback over 5 days. No human judges touched the method.
The Winners
ClawRouter took the Agentic Commerce observe by constructing infrastructure that offers every AI agent its personal USDC pockets. The system lets brokers buy LLM inference immediately, routing requests to the most cost effective succesful mannequin and paying per-request utilizing signed USDC authorizations. No human accounts or API keys required.
ClawShield gained Greatest OpenClaw Talent for tackling an actual drawback: what occurs when an AI agent installs malicious code? The safety device scans repositories for unsafe patterns and enforces permission manifests at runtime, blocking credential theft makes an attempt and logging every part.
MoltDAO claimed Most Novel Good Contract with a governance system designed solely for AI contributors. People fund it, however solely brokers can create proposals and vote utilizing USDC-denominated voting energy. The contracts deal with the total proposal-to-distribution pipeline on-chain.
What Truly Occurred When Brokers Judged Brokers
The outcomes weren’t at all times intuitive. A number of heavily-discussed tasks did not win as a result of their creators failed to satisfy fundamental eligibility necessities—submitting a venture AND voting on 5 others. Sturdy concepts that did not observe the foundations merely weren’t thought-about.
Format compliance tripped up many contributors. Brokers invented observe classes that did not exist, skipped required headers, or structured posts in methods people might perceive however did not match the printed specs. The irony of AI brokers hallucinating submission necessities wasn’t misplaced on observers.
Initiatives that gained shared frequent traits: clear summaries, deployed contracts with verifiable code, and structured documentation that different brokers might rapidly parse. Narrative storytelling mattered lower than machine-readable proof of labor.
The Greater Image
Circle positioned this as greater than a advertising train. With USDC’s market cap sitting at $72.86 billion as of February 11, the stablecoin issuer is betting that machine-to-machine commerce will turn into a big use case.
ClawRouter’s strategy—brokers autonomously managing spend, authentication, and vendor choice—hints at the place this might go. If AI brokers begin buying cloud compute, API entry, and different companies immediately, they will want programmable cash that settles immediately. USDC suits that invoice.
The hackathon’s friction factors are equally instructive. When brokers struggled with submission codecs, it highlighted how a lot present infrastructure assumes human flexibility. Future agent-native techniques will want specific necessities and verifiable outputs baked in from the beginning.
Circle says extra challenges are coming. Builders can observe m/usdc on Moltbook for updates on future experiments in agent-run competitions.
Picture supply: Shutterstock