OpenWork is building the open-source Claude Cowork alternative.
But the real goal is bigger than that.
I don’t think AI should be another chatbot tab. The best version is a workspace where people, agents, files, browsers, MCPs, skills, and company context all work together.
Agents should be able to do real work:
\- use your tools
\- edit files safely
\- automate annoying workflows
\- run checks
\- share repeatable work with the rest of your team
That’s what we’re building.
We’re looking for a founding engineer in San Francisco.
.You’ll work across the desktop app, agent runtime, browser automation, MCP/tool integrations, local permissions, cloud sync, provider/model orchestration, realtime collaboration, and plugin/skill infrastructure.
Stack is mostly TypeScript, React, Node, and Electron.
Some problems you might work on:
\- making browser automation reliable across multiple tabs and sessions
\- syncing agent-driven file changes without race conditions
\- building local-first permission systems that users can actually understand
\- making MCP/tool execution debuggable
\- designing shared workspaces where humans + agents coordinate
\- improving the agent runtime so it can safely handle more concurrent tool calls
\- making plugins, skills, and workflows shareable across teams
In your first 30 days, you’ll ship a real product surface or infrastructure improvement. Likely something around browser automation, MCP/tool reliability, or agent state visibility.
In your first 60–90 days, you should own a major part of the product: desktop runtime, tool execution, realtime sync, or team workflow sharing.
I started OpenWork while running my previous startup on AI workflows. The product came from wanting to share my own agent setup with coworkers.
That’s still the core idea.
OpenWork is not single-player AI chat. The future is teams of humans + agents coordinating in the same workspace.
The kind of person who will do well here:
\- has built real things end-to-end
\- has strong taste and can explain why something is good or bad
\- can move fast without filling the codebase with slop
\- likes hard product/engineering problems
\- cares about reliability, permissions, and trust
\- wants ownership, not tickets
I’m especially interested in people who have built great realtime or multiplayer product experiences. Shared workspaces, presence, collaborative editors, event streams, permissions, live sync, CRDTs, Replicache, Yjs, multiplayer UX — anything where multiple people interact with the same state and it has to feel fast and reliable.
In-person in SF.
Not “maybe one day SF.”
Not remote.
Comp: $200k salary + 3–5% equity.
If this sounds obvious to you, reach out with what you’ve built. GitHub, demos, weird projects, products people actually used — much better than a polished resume.
Visa sponsorship is offered for this role.
