I've been a fan of OpenClaw since its ClawBot days, and found the various clones/ports to be limited, but this project is genuinely interesting: a personal assistant designed for local inference and efficiency in mind. It has a great team behind it: https://scalingintelligence.stanford.edu/blogs/openjarvis/ I've been wanting to explore OpenHands + OpenClaw, but I may start with OpenHands + OpenJarvis instead 😄

I'm curious to hear your thoughts on how and why to couple openhands with openclaw? What use cases or scenarios do you think openhands on it's own is sufficient and where do you feel openclaw needed that openhands alone wouldn't suffice? I don't know enough about either projects so appreciate your sharing your learnings

OpenClaw has taken the world by storm. My team has published 10 articles on it on our tech blog in the last month. It has had exceptionally fast growth, taking only four months to become the highest-starred software project - surpassing React and the Linux kernel which took 10 and 30 years respectively to reach a similar number of stars. The main value proposition of OpenClaw is many-fold: • a personal agent that lives in an environment it can interact with, installing software, setting up services, configuring and updating itself, creating skills on the fly • connecting to various sources of public, organization, and personal data: e.g. web search, Google Workspace, emails, etc. • meeting you "where you are" across various messaging platforms: WhatsApp, Discord, Telegram, iMessage, Slack, and even webchat interfaces • Having natural language chat as the UX for agent development, execution, and management. OpenHands covers a bunch of this, but the viral sensation has created a vibrant community, tools and skill ecosystem, derivative ecosystems, etc. Also the low/no friction (there are many 1-click deploy OpenClaw solutions from Kimi, MiniMax, AWS, Genspark, etc.), auto-setup / auto-config / self-upgrade, and conversational UX in OpenClaw are hard to beat. That's also why there have been many incumbents trying to fork/port/copy it but they have not gained any traction (e.g. nanoclaw, zeroclaw, picoclaw, zclaw, etc.). OpenJarvis seems to be an exception to that because it's not a one-person vibe-coded project. There are notable people and organizations behind it. And it's designed around local inference and efficiency as a first class principle. That's why I'm interested in giving it serious consideration. As far as how OpenHands could integrate with OpenClaw (and/or OpenJarvis): There are many ways, several of which still need to be thought of and explored. One immediate path is using the recent ACP Agents integration, to fire off OpenHands subagent sessions and feed info back to the chat via ACP. Another is to have a conversational interface for multi-session management and insights, like the app control plane idea that was discussed last month, but with chat UX.