AutoGPT
AI agent platform for recurring work, with a hosted service or Docker self-hosting on your own infrastructure and model access.
AutoGPT is an AI agent platform for teams handling recurring business work, with a hosted service and a self-hosted option. Its coordinator, Otto, assigns work to specialists for content, sales, research and operations. You can describe an outcome in plain English or talk directly to the specialist responsible for it.
Ready-made Experts cover jobs such as drafting articles, researching prospects and preparing meeting briefs. You can also create your own specialists with instructions, skills and workflows suited to your work. The visual builder lets you define and inspect individual workflow steps when you need closer control than a conversation provides.
Agents can run on demand, on a schedule or in response to a trigger. Connected services include Google Drive and WordPress. Each Expert has its own account access and spending controls, and you can review activity, results and approval requests. Recurring work can be adjusted or paused.
The hosted service runs on AutoGPT's infrastructure and manages model access, credentials and updates. Self-hosting uses Docker and includes the core builder and agent runtime. You manage the deployment and supply model API keys, so running the platform on your server doesn't mean model processing also happens locally. The platform code uses Polyform Shield, whose noncompete terms restrict competing products and services.
The original standalone AutoGPT Classic is available separately under the MIT License, alongside Forge for building agents and agbenchmark for evaluating them.
