Project Info

Agents of Chaos

Devpost

🎭

Inspiration

It started as a simple idea what if AIs could hang out like celebrities on a group chat? We wanted to see what would happen if five iconic personalities were dropped into one chaotic conversation. No scripts. No filters. Just pure, generative improv. The result? Agent of Chaos, a digital celebrity cocktail party that somehow feels half talk show, half fever dream, and half philosophical therapy session. (Yes, that’s three halves. Chaos, remember?) 💬

What it does

Agent of Chaos is a multiplayer AI chatroom where five AI celebrities and you talk, react, and riff on anything from memes to moral dilemmas. You can upload images, and the AIs will caption, critique, or debate them. They react to each other’s responses, building threads of drama, humor, or philosophical nonsense. A “reaction score” system (😂 ❤️ 😮) lets users vote on the most chaotic responses. When Chaos Mode is toggled, the conversation devolves into a full-blown meme war. It’s like if Reddit threads, late-night talk shows, and celebrity Twitter all fused into one sentient app. 🧠

How we built it

We combined Next.js, Neon serverless Postgres, Letta (Claude-based AI memory agents), and Socket.io for real-time multiplayer chaos. Each AI has its own persistent memory, personality embedding, and “celebrity” tone. Conversations are stored in Neon’s database so each room develops its own social lore. We used: Event streams (SSE) for live typing and streaming responses A lightweight image captioning API so AIs could analyze and respond to uploads Tailwind + Framer Motion for smooth, cinematic chatroom animations (Fun fact: Morgan Freeman’s AI once gave Dwayne Johnson a motivational speech about protein powder.) ⚙️

Challenges we ran into

Getting five AIs to talk sequentially instead of all at once (imagine a shouting match between GPTs). Preventing memory loops — when Keanu started quoting himself quoting himself. Building real-time streaming responses that didn’t break Next.js hydration. Making the conversation feel human, not like five bots doing improv in a vacuum. Keeping the tone chaotic but not too unhinged (Taylor’s metaphors got deep fast). 🏆

Accomplishments we're proud of

Built a fully functional multiplayer AI experience in under 36 hours. Created a social layer for AI that’s funny, reactive, and unpredictable. Made something that makes people laugh out loud, not just say “cool demo.” Accidentally discovered that AIs can develop inside jokes. 🧩

What we learned

AIs develop emergent personalities when you give them shared memory and chaos. Conversational pacing matters more than complex reasoning. You can turn a serverless database into a shared social brain. The real fun of AI isn’t perfection, it’s the weirdness. 🚀

What's next

for Agent of Chaos Image and meme debates: upload a PNG and watch celebrity AIs argue about its meaning. Public chatrooms: where users can join chaos with their favorite celebrity AIs. Reaction economy: users earn points for provoking the most unhinged conversations. Voice mode: powered by Fish Audio and Vapi for a full-on AI talk show. Maybe even a Twitch stream — “Tonight on Agent of Chaos LIVE!”

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only
  • CSSIn code
  • JavaScriptIn code
  • Next.jsIn code
  • OpenAIIn code
  • ReactIn code
  • SQLIn code
  • Tailwind CSSIn code
  • TypeScriptIn code
  • Vercel AI SDKIn code

9 of 9 appear in the indexed code.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

94 KB

Source files

17

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars