AIE Code Summit 2026 Hackathon

Labbit is a minimal live robotic-hand workbench and a reproducible simulation dataset factory. The goal is 100,000 useful, diverse 20-second episodes of…

A lot of what you send a coding agent is easy - “yes, push it”, “commit it”. Sending that to a cheap model sounds smart, but every switch kills the prompt…

Frankenstein brings back the lost source code of classic games by rewriting it, byte-for-byte. Meta's Muse reads a game's compiled cartridge and rewrites the…

Self improving editing software for short form content. "one must imagine Sisyphus editing"

Clean Room rebuilds software by watching it. No source code, no decompiler, no peeking under the hood. You show it an app running, a phone calculator, a…

child-1 makes LLM decode faster on the Qualcomm Hexagon NPU and proves every speedup on a real phone. It optimizes llama.cpp's Hexagon NPU backend…

Factory failures get closed on a yield bump. Then the defect comes back. Closeout makes a fix earn its closeout. We built an agent-assisted failure analysis…

EvoArena: Lantern Hall turns software archaeology into a playable mystery. Claude and Meta’s Muse Spark independently explore a world of light, glass, crates…

ClauseLens turns a stack of signed contracts into answers a reviewer can trust and act on. Most companies have hundreds of signed agreements and no clear…

Run many agents to see how changing weights to fix a whisper small model can fix it and gain learning from it

Game Archaeologist turns a compiled game into readable, editable code without using its source. Meta's Muse Spark investigates the ROM through disassembly…

Chatmate — Trust but Verify Chatmate is an independent verifier for AI-generated code, available as (1) a VS Code extension with a separate verification…

Adversary: an independent verifier for AI-written code (Trust but Verify). AI writes code faster than teams can review it, and when one model writes both the…

Your coding agent says the patch is ready. Every test is green. What did those tests miss? CrossExam challenges the tests themselves. Its architecture…

Thought to Reality turns a natural-language request into a mechanical part or a full product. A designer model writes parametric CAD; an independent critic…

Capy Kondo turns one photo or a 6-second video of the whole pile into honest, priced listings in about a minute: Meta Muse Spark and SAM 3.1 find and cut out…

partial reconstructs a damaged biosignal recording from its raw serial tape and says which parts it saw and which parts it guessed. A Cerelog research board…

AI coding agents love to say “Done, all tests pass.” I’ve spent five years building finance integrations, where code that looks done but isn’t moves real…

Strata — an AI archaeologist for software whose source is lost. Software outlives its code. Games lose their servers, devices die with the company that made…

AI writes more code than anyone can read. We treat code as a black box: settle the inputs, outputs and tests up front, then verify the result against them.

NeverTwice is a self-evolving pre-merge referee and episodic memory engine that stops autonomous AI coding agents from shipping buggy code and repeating past…

amusingkernels is an LLM-driven agent that writes, debugs, and optimizes GPU kernels for NVIDIA's SOL-ExecBench benchmark, producing leaderboard-ready…

Theseus is a human-facing control plane for an AI-native organization. Such an organization never stops evolving: hundreds of agents and people change how the…

Dojo trains a robot skill from one sentence. Type “put the red block in the green bowl” and it builds a MuJoCo scene for an SO-101 arm, collects expert demos…

Describe a broken thing. Get a part that physics has checked. Prompt-to-mechanical-part: an agent designs a 3D-printable replacement part, an independent…

Scar Tissue — The Coding Agent That Forgets What Doesn't Help Most coding agents never get smarter. They just accumulate: every reflection saved, every…

Puddles makes coding agents better at working your way, so you don’t have to repeat the same corrections. Built as a fork of T3 Code, an open-source workspace…

We built a self-improving agent harness for finance. The agent runs on the Pi coding-agent runtime (sandboxed, with read/bash/write tools) and improves by…

Hidden-test gating catches overfit harness mutations in a frozen coding LLM (Meta muse-spark-1.1; 16 tasks; 369 independently graded runs). 3/3 grader-caught…

Rebrowse is a Chrome extension that learns skills by observing how you use the web, combining your live demonstrations with days' worth of browser automation…

AI agentic harness for training FM models

AI that overturns your migration plan with proof, not opinions. Compat Archaeologist decides whether you can leave MongoDB on evidence, not paper. It…

Learned robots work in most cases but fail on niche edge cases. Their decisions live inside millions of neural-network weights, so teams can’t tell exactly…

AI writes code faster than teams can check it. PRProof is an independent verifier for AI-written pull requests: instead of asking you to trust the code or the…

Independent Code Reveiwer for AI Code for Cybersecurity and Customer Experience Sharley helps code repos, code ideas get production ready by finding buggy…

Picture a powder morning at Palisades Tahoe. Hundreds of people stand in a line that is not moving, because the Tram and the Funitel, the two gate lifts the…

Market Claim Verifier checks the numbers in finance posts before they ship. Paste a post: Meta's Muse Spark (Meta Model API, also Baseten and Fireworks)…

Doppel catches bugs in AI-written code, with proof. Paste a function and what it should do. Meta and Google each rebuild it blind from the description…

We built Bob, a multi-agent system that helps AI coding agents improve safely. Bob analyses an agent’s failures, proposes targeted configuration changes, and…

Colony: a coding swarm that learns from every fall Autonomous coding agents stall on long, tightly coupled tasks because one wrong assumption early, like…

Kedge. Live demo: https://kedge-demo-steel.vercel.app/improve https://kedge-demo-steel.vercel.app/ A coding system that gets better at porting legacy…

We use RSI to automatically improve a language model's physics: builds the training environment, data, and checker-based reward, then fine tune a Qwen model…

The problem - Supermarket cold rooms run on old controllers that accept unauthenticated commands. In the Oct/Nov 2026 EU attacks, a single remote write could…

Merge Gate is an independent verifier for AI-written pull requests. It never trusts the PR's own tests or description. Instead, Meta's Muse model writes…

Autonomous multi-agent framework that scans codebases via a swarm of specialized agents to fix bugs, optimize system design, UI/UX, and cloud costs. It…

Doppel is an AI agent framework you can trust: an agent that does real work, plus a verifier that checks every step before anything ships. Meta's Muse models…

Self-optimizing trading systems

CodeJury — Independent AI Code Verification AI writes code faster than teams can review. CodeJury is a jury of Fetch.ai agents that independently verify…

Second Grader: trust, but verify your coding agent Coding agents describe their own work, and that description is often wrong. An agent will say "all tests…

Second Opinion is a multi-agent system for writing code you can trust. When AI writes code and also tests it, it tends to agree with itself, so bugs slip…

this mission requires the Robotic Dog walkaround a certain area of a Mars surface and return to the base/spaceship; record each route attempt and save it as…

Imagine giving an AI agent access to your money and telling it to make trades on your behalf. You trust it to make decisions, but what happens when it makes a…

Resurrect rebuilds old software from a short screen recording, without the builder ever seeing the original app, and proves the new version behaves the…

Every day, more than 100,000 new vibe-coded projects emerge, App Store submissions are up 104% vs 2025, and yet AI coding is barely getting started. But this…

FOUNDRY — Build. Prove. Evolve. AI can build agents in seconds. But who proves they work? And what if the AI engineer could learn from its own…

ReleaseProof Recovery by AIZOYA Independent evidence before a software repair advances. A repair can look convincing and pass one visible example while still…

Trucking companies connect with their Freight Forwarder clients via API. However, when there are communication issues, who can help resolve and bring a…

AdaptMach is an autonomous machine-design agent. You describe a machine, like “deliver a pizza from San Francisco to Fremont through a hurricane”. The agent…

AI-generated code can look correct while hiding subtle bugs that are easy to miss in review. ProofHunter helps maintainers check whether a pull request breaks…

ReflexQA is a recursive, self-improving code QA agent. The speed at which we can push code has exponentially increased. The speed at which we can review code…