Auto Learning Agents

Auto Learning Agents is a free open source operating system for AI agents. It runs many agents in parallel, each with its own schedule, model and instructions, sharing persistent memory and learning from outcomes through a web UI.

Auto Learning Agents is a free, open source operating system for AI agents, published by the developer behind the AI Apps API network. The idea is to run any number of agents side by side, with each agent carrying its own schedule, its own model choice and its own instructions. A full web interface sits on top, so agents are configured and watched from the browser rather than through scattered scripts. Two traits set it apart from a basic agent runner. Agents share a persistent memory, so what one learns can inform the others, and the system learns from outcomes, meaning results of earlier runs feed back into later behavior. That makes it closer to a managed workforce than a one-off automation, which is how the creator describes it: the nearest thing to hiring a team that costs nothing. The intended audience is developers, solo operators and small teams who are comfortable with open source software and want control over their own agent setup instead of a hosted subscription. The code lives on GitHub, and the project has its own site at autolearningagents.com. The homepage describes it as the flagship of thirteen free open source projects built by one developer, and it is the item the author singles out if a visitor only looks at one thing. In practice, a user would define an agent, assign it a model and instructions, set when it should run, and let the platform execute several agents in parallel while memory accumulates between runs. Related projects from the same author cover adjacent needs, including WebBrowserBot for browser automation and Adaptive Recall for a hosted memory API over MCP and REST. Among alternatives, it competes with no-code agent builders and workflow automation services on one side, and with agent frameworks that require writing orchestration code on the other. Its position is a self-managed, web-driven platform with built-in scheduling and memory, free to use. Buyers should note that it comes from a single maintainer, and the source gives no details on hosting requirements, supported models or formal support.