Crew, an multiplayer workspace for humans and AI agents to work together
My brother and I built Crew because we were tired of jumping between separate AI chats, terminals, file explorers, design tools, and collaboration apps just to build one project.
Crew is a shared workspace where you, your teammates, and multiple AI agents can work on the same project at the same time.
In the video, I’m showing a few parts of it:
- Multiple LLMs and agents working simultaneously, even inside the same thread
- Agents creating, editing, viewing, and deleting project files
- A built-in terminal with full CLI access
- Sub-agents that can split larger tasks into smaller pieces and work in parallel
- A design workspace where an agent can design live in front of you
- The ability to edit the design alongside the agent while it is still working
- Huddles where everyone in the workspace can talk together, including your AI agents
- File previews, Git tools, project boards, shared memories, plans, and usage tracking
The design workspace is probably the part I’m most excited to show.
The agent doesn’t just generate a finished screenshot and hand it to you. You can actually watch it place, move, and edit elements like a human working in Figma. At any point, you can jump in and continue designing alongside it in real time.
We created Crew’s interface and design ourselves. We also didn’t clone an existing GitHub project, reskin it, and call it a design feature. We designed and built it hands-on from the ground up because we wanted Crew to feel like its own product.
We added sound effects and small interactions throughout the app because we wanted Crew to be fun to use. We didn’t want to make another lifeless AI project where everything feels the same. We genuinely just wanted to have fun building it.
I’m posting this here because I know a lot of people in this community already use several agents at once, while others enjoy designing just as much as coding. Crew brings both of those workflows into one place. You shouldn’t need to jump between five apps or pay for a separate Figma subscription just to sketch and design with your agents.
Crew is fully open source, and the entire application runs on your computer. It works with the LLM CLI tools you are already signed into, so there are no separate API keys or additional model subscriptions required. You bring your own agents, and Crew detects the supported LLM CLIs installed on your machine.
You can also install the Crew terminal command from Settings → This Computer. After that, you can run crew inside a project to open that folder directly in Crew.
We’ve also been working on Scribe, our completely local speech-to-text feature. Scribe runs entirely on your computer, requires no API key, and is completely free.
Scribe can also understand links while you speak. For example, I could say, “my GitHub is https://github.com/alihammoud21,” and Scribe can turn that part of the transcription into a clickable link instead of leaving it as plain text.
We’re also working on deeper iOS development support. You can already use Crew to work on iOS projects, but we want to bring simulators directly into the workspace so you can build, run, view, and interact with an iOS app without constantly switching windows.
The idea is similar to the iOS Simulator workflow Claude Code recently introduced, but built around Crew’s multiplayer and multi-agent workspace.
We even added synchronized agent-generated music as a small experiment, so you can vibe to music while you vibe code. We may have gotten a little carried away with that one lol.
We’re still building and experimenting, so honest feedback would mean a lot. I’m especially interested in what you would use Crew for and what you think is still missing.
GitHub: https://github.com/JamelHammoud/crew
This entire description was spoken and transcribed using Scribe, running completely locally on my MacBook.
Made with love ❤️ by Ali and Jamel.
