BabyAGI is an open-source autonomous agent project posted to GitHub in late March 2023 by Yohei Nakajima, a venture capitalist and independent AI researcher who serves as General Partner at Untapped Capital. Released as under 200 lines of Python, it became one of the founding reference implementations of the autonomous-agent pattern alongside AutoGPT, and its repository passed 12,000 GitHub stars shortly after release.
Given a single objective, BabyAGI runs a loop of three specialized agents: one executes the current task with an LLM, one generates new follow-up tasks from the result, and one re-prioritizes the task list before sending the next item back to the execution agent. Later variants, including babyagi-2o, extended the original concept toward a more general self-building agent framework.
BabyAGI is aimed at developers and researchers experimenting with autonomous agent architectures, not at non-technical users looking for a hosted app. It is free and open source under Nakajima's GitHub account, needing only a separately billed LLM API key to run.
