September 2026
Referee
A live hackathon judge: Devin clones, deploys and scores every team's repo in its own sandbox.
All 11 recent builds, newest first
11 projects
September 2026
A live hackathon judge: Devin clones, deploys and scores every team's repo in its own sandbox.
September 2026
One endpoint for every model, with a router that learns from LLM-judge scores which model to use.
September 2026
A deliberately flawed HR agent and the four layers of evals and red-team tests that catch it. Attack success went from 86% to 0%.
July 2026
An AI editor that ingests, deduplicates and clusters the news into stories, with briefings and chat.
June 2026
A phone voice agent that recommends vinyl and reserves records while you're still on the line. Won the one-hour hackathon.
June 2026
Voice AI intake and dispatch for property restoration teams that files tickets while the caller talks.
June 2026
A white-label, real-time, play-money prediction market where AI forecasters trade alongside people.
June 2026
An agentic RAG navigator for the EU AI Act that runs locally on a CPU and checks every citation. 97.9% faithfulness, 233 tests.
May 2026
An AI agent on AWS that pays for paywalled APIs on its own, a cent per call in USDC, over x402.
May 2026
Market intelligence where four agents turn raw market signals into findings for specific roles. 1st place out of 105 teams.
January 2026
Gamified analytics for a team's Cursor usage, with leaderboards, achievements and an AI chat. Used by 40+ developers.