LangChain
Agent engineering platform for building LLM applications from composable parts
Best for: people who want a personal AI assistant that keeps a local memory of their work and can orchestrate other agents.
OpenHuman describes itself as a personal AI super intelligence: a local-first brain that builds memory of your life, an orchestrator for agent fleets and workflows, and a deep researcher. It is written in Rust and licensed under GPL v3, with community activity on Discord, Reddit and X.
What makes it interesting: local-first memory. Where most assistants forget everything between sessions or store your history on someone else's servers, OpenHuman is built around retaining context on your own machine, combined with orchestration so it can dispatch work to multiple agents rather than doing everything in one conversation.
Where it falls short: the maintainers label it early beta and under active development, and tell you to expect rough edges — take that at face value before depending on it. They are also refreshingly direct that OpenHuman is not AGI, despite the framing, describing it instead as an architectural step toward better memory and orchestration. GPL v3 has real implications if you plan to build commercially on top.
Details: Rust, GPL 3.0. Early beta.
Agent engineering platform for building LLM applications from composable parts
Agent harness optimisation for Claude Code, Codex, OpenCode and Cursor
Self-improving agent from Nous Research that learns across sessions
Small, composable agent skills from Matt Pocock's daily .agents directory