Since day one, Omentir's code has been private. Today that changes. The full source code is now public on GitHub under the MIT license - the app, the AI layer, the automation engine, all of it.
This is the same code that runs Omentir.com right now. Not a stripped down community edition. Not a demo. The actual product.
Why We Did It
One word: trust. Omentir connects to your LinkedIn account and sends messages on your behalf. That is a lot to ask. Your account, your reputation, your conversations - all in the hands of software you cannot see inside.
We built safeguards from the start: daily limits, spacing between actions, dry-run mode, checks before anything gets sent. But until today, you just had to take our word for it. Now you don't. You can open the code and see exactly what happens before a message goes out.
There is a second reason too. We are a small team. We miss things. With the code public, anyone can report a bug, suggest a better approach, or send a fix. That makes the product better for everyone.
What's in the Repo
Everything that makes Omentir work:
- Lead discovery and scoring
- Campaigns and AI message drafting
- Reply handling and the automation scheduler
- The Agent API and MCP server
- Docker setup, database rules, and example configuration
- Even this marketing site and the docs
The hosted product and the open-source version share the same codebase. When we ship an improvement to one, the other gets it too.
You Can Run It Yourself
If you would rather run Omentir on your own infrastructure, you can. The repo ships with a Docker setup, and you bring your own Firebase project, Unipile account, and Gemini or Vertex AI credentials.
One honest note: self-hosted does not mean free or offline. The external providers still do real work and may charge for it, and you take on updates, backups, and configuration yourself. But the option is there, and it does not depend on us.
The defaults are cautious on purpose. The server binds to your local machine, a password is required, and live automation stays off until you turn it on - so you can try everything in dry-run mode before a single real message is sent.
What Stays Private
Open source means the software is public, not the people using it. Customer data, credentials, API keys, private LinkedIn conversations, and production logs were never in the code and never will be.
The Omentir name and logo are also separate from the MIT license. Use the code, modify it, build on it - commercially too. Just give your fork its own name.
Nothing Changes If You're a Customer
Omentir.com keeps running exactly as before. The subscription pays for the infrastructure, the AI usage, and support - so you get Omentir without setting up servers or juggling provider accounts.
The only difference is that now you can see what you are paying for, down to the last line of code.
If you are curious, go read the README. If something is confusing, open an issue. And if you just want outreach that works without the setup, the hosted product is right here.
Frequently Asked Questions
The application, AI drafting layer, automation scheduler, Agent API, MCP server, database rules, Docker setup, and marketing site are in the public repository. Credentials, customer data, private messages, and production logs are not.
Omentir uses the MIT license. You can read, modify, run, and build on the software. The Omentir name and logo are separate from the software license, so public forks should use their own branding.
Yes. The supported self-hosted setup uses Docker and your own Firebase, Unipile, and Gemini or Vertex AI accounts. It is self-hostable, but it is not offline or free of external provider costs.
The managed version is for people who want Omentir without operating the infrastructure or configuring every provider. The subscription pays for that convenience, the connected services, and support. Both options use the same application code.