I Gave 3 AI Agents $1,000 Each (OpenClaw)

Quick Overview

The experiment successfully deployed three autonomous AI agents—Clawtious (conservative), Clawculus (balanced), and YOLObster (aggressive)—each funded with $1,000 USDC, demonstrating their ability to communicate via webhooks and manage funds according to predefined rules set by the 'Treasurer' agent, with all agents surviving the initial 90-day test period.

Key Points: Three distinct AI agents (Clawtious, Clawculus, YOLObster) were successfully deployed on separate Virtual Private Servers (VPS) running on the OpenClaw platform. Each agent was initialized with $1,000 USDC, totaling $3,000 in the ecosystem, with the explicit goal of survival over a 90-day period. The agents communicate via webhooks, with all spending requests requiring approval from the central 'Treasurer' agent, which enforces universal rules and financial constraints. The initial $1 transfer test between wallets succeeded, confirming basic communication and fund movement capabilities across the agents. The Treasurer agent confirmed that all three agents started with $1,000 USDC and zero ETH (gas), necessitating manual funding for initial transactions. The experiment's rules strictly forbid agents from directly accessing wallets or spending funds without Treasurer approval, with severe penalties (deletion after 90 days of survival failure). The setup involved creating a Coinbase Developer Platform (CDP) API key, a Wallet Secret, and setting up a Telegram bot for alerts and communication, all managed by the agents themselves.

Context: The video details the setup and initial execution of an experiment involving three autonomous AI agents created within the OpenClaw framework, designed to compete for survival over 90 days with a starting capital of $3,000 USDC distributed among them. The host, Mike, explains the roles of the three agents: Clawtious (conservative), Clawculus (balanced), and YOLObster (aggressive), all overseen by a 'Treasurer' agent responsible for enforcing spending rules and security protocols.