Skip to Main Content

Car1y 🚗

Built at The Recursive Agents Hackathon · Oct 17, 2026 · San Francisco, CA

Car1y 🚗 — Demo video

Carly: an AI agent team that buys and sells used cars for a dealership, and gets better with every deal. Used-car dealers spend all day scrolling Facebook Marketplace, Craigslist and Autotrader, checking prices against Kelley Blue Book (KBB), and haggling with sellers and buyers. Whatever they learn about a seller, a model or a tactic stays in one person's head. Carly runs that work on the dealer's own machine, and keeps running after the staff go home: Buy side: finds live listings through You.com, appraises each car with cited price data, and negotiates with sellers. It never offers above the dealer's maximum. Sell side: writes one listing and posts a version to each channel, answers buyer questions, and closes above the dealer's minimum price. The dealer approves every purchase, and any sale below the minimum price. The code enforces this; it isn't just a prompt instruction. Shared brain: every agent reads one shared memory before it works and writes to it afterward. That memory holds the dealer's rules, a profile for each seller and buyer, market prices, and a win rate for each negotiation tactic. Agents only add entries and never overwrite each other, so their writes can't conflict. Hard size limits keep the memory small. Each agent also has its own list of past mistakes, so a scam link spotted on Facebook won't fool that agent again.

Team