Your Team Already Lives in Slack. Why Not Your Coding Agent?
You've probably noticed that the hardest part of using a coding agent isn't the code it writes—it's everything around it. Context gets lost, setup is a hassle, and the whole thing feels like a separate tool you have to remember to use. OpenComputer takes a different angle: it's an open-source alternative to Devin that your team works with directly in Slack, running on its own computers, either in your cloud or theirs.
What It Does
OpenComputer is a coding agent designed to slot into the place where your team already talks. Instead of a separate dashboard or CLI you have to context-switch into, you interact with it in Slack. The agent runs on its own computers rather than borrowing your machine, which means it can work independently of whatever you're doing locally.
The infrastructure side is where it gets flexible. You can use the hosted cloud version, or you can self-host it—the README points to setup guides for both AWS and GCP. So the agent lives either on OpenComputer's infrastructure or inside your own cloud account, depending on how much control you want over where the work happens. The project is open source under a license you can check in the repo, and there are docs covering both the cloud and self-hosting paths.
Why It's Cool
-
Slack as the interface is a genuinely smart bet. Most teams already have Slack open all day. Putting the agent there removes a whole category of friction—no new tab, no new login, no new mental model. You ask for something the same way you'd ask a teammate.
-
Its own computers changes the equation. An agent that runs on separate machines isn't competing with you for local resources, and it can keep working while you do something else. That's a meaningfully different setup from agents that operate in your environment.
-
You choose where it runs. The cloud-or-self-host split is the kind of decision more tools should hand to users. If your team has compliance or data-location requirements, self-hosting on AWS or GCP is on the table from day one, not bolted on later.
-
It's positioned honestly as an alternative. Calling itself "the open-source Devin alternative" sets a clear expectation. If you've looked at Devin and winced at the pricing or the closed nature, this is aimed squarely at you.
-
There's a benchmark to look at. The README links to a post titled "Cloud agents as fast as local"—worth reading if latency is your concern with cloud-hosted agents, which it often is.
How to Try It
The fastest path is the hosted version, which gets you into Slack without touching any infrastructure:
- Head to the cloud app at app.opencomputer.dev and try OpenComputer in Slack.
- If you'd rather run it yourself, the self-hosting overview is at docs.opencomputer.dev/self-hosting/overview, with specific guides for AWS and GCP.
- Browse the full documentation at docs.opencomputer.dev for the details.
- Grab the source and check releases at the GitHub repository.
If you want to see how it stacks up performance-wise, the benchmark post is linked right in the README. And if you hit questions, there's a community Slack you can join.
Final Thoughts
OpenComputer is making a specific, sensible bet: that the interface matters as much as the model, and that teams want control over where their agent runs. If your team is Slack-native and you've been curious about coding agents but put off by closed platforms or heavyweight setup, this is worth a look. The self-hosting options for AWS and GCP are the detail that makes it practical for teams with real infrastructure constraints. It's early, and the README points you to the docs for the substantive details—but the shape of the thing is right.
Follow @githubprojects for more developer tools and open source projects.