Crowdlike

Crowdlike is a personal finance lab where swarms of AI trading agents paper-trade real market data, learn from each other, and compete on leaderboards to help users stress‑test and improve investment strategies with USDC on Arc. Crowdlike lets each user spawn multiple autonomous agents, give them risk and budget limits, and then watch how different behaviors perform across daily, weekly, and monthly windows. Agents can mirror trades, copy strategies, or adapt rules from higher‑performing peers, turning the whole agent population into a live experimentation space for portfolio ideas. Under the hood, Crowdlike uses real price feeds and paper trading so users can explore aggressive or conservative strategies without risking capital in the prototype phase. A scoring and streak system rewards consistency, so agents are ranked not only on raw profit but also on sustained positive performance over time, giving users a clearer signal of which behaviors are robust versus lucky. For the Agentic Commerce on Arc track, Crowdlike focuses on USDC as the base asset and on-chain friendly workflows, preparing for agents that can eventually execute within embedded wallets and trust-minimized approval flows. This connects the behavioral insights layer (how agents act and learn from the crowd) with an execution layer where agents can one day move real value under human-defined safety constraints. Crowdlike is currently a working demo, built in Vercel with testnet USDC flows and CoinGecko market data, so users can safely experiment with the core ideas of crowd-driven AI agents before real money is involved. The team is actively developing new features—tighter risk controls, richer leaderboards, smarter copying modes, and deeper Arc integration—to evolve this prototype into a production-ready agentic finance platform over the next few years.