Most "AI agent" products are framework demos with a pricing page bolted on. You stitch a prompt to a tool, wrap it in a loop, ship it as a SaaS, and call the result an autonomous worker. It is not autonomous and it is not a worker. It is a prompt with a UI.
Team-X starts from a different first principle: if you are going to operate AI agents at any real scale, you are not running a prompt. You are running a company. A company has a hierarchy. It has roles, not assignments. It has tickets, not chat threads. It has meetings, audit logs, schedules, KPIs, and a chain of decision authority. It has a place where the boss can ask "why is the frontend team behind schedule" and get a grounded answer that names specific tickets and specific employees.
Team-X is the operating system for that company. Local-first by default. Zero phone-home. Open under MIT on commit number one. No upgrade tax. No proprietary cloud. The product on your machine is the same product on every other machine.
The product on your machine is the same product on every other machine.
What it is not
It is not a hosted SaaS. It does not run in your browser tab. It does not require an account, an email address, or a credit card to run end-to-end. It does not collect telemetry. It does not phone home. There is no "team plan" with extra features locked behind a per-seat fee.
It is not a framework. You don't write code to compose agents; you hire from a curated library of 57 hand-written F10 role specs and you build an org chart by dragging cards. You file tickets and assign them. You call all-hands meetings. The agentic surface is in Cmd+K, not in your IDE. The framework is the chain of command.
It is not a content marketing site for an enterprise sales motion that doesn't exist yet. The repository is the spec. The release page is the pricing page. The newsletter (when it ships) will be operating notes from running our own AI org, not landing-page filler.
Who I am
I am Rocky Elsalaymeh . I run Strategia-X , a small consulting firm that builds, ships, and operates technology for clients who care about the difference between working software and software that demos well. Team-X is one of several open-source products that come out of our shop.
I do not work in venture-backed AI. I do not run a Slack with a hundred employees. I am not selling you a course. The reason Team-X looks the way it does is that I built the first version of it for myself, to operate the AI side of my own work, and the version you can download is what I run on my own machine every day.
Why open-source on commit number one
Open-source on day one is the only honest version of the product. If the privacy posture says your data stays on your machine and the source is the spec, then anyone has to be able to read the source to verify the spec. There is no separate "enterprise edition" with the privacy plumbing locked away. The MIT license is the contract; the repository is the proof.
It also raises the bar on the work. You cannot ship a half-finished feature with a TODO and a screenshot when the issue tracker is public. You cannot dress up a flaky integration with a marketing page when there's a Playwright spec that has to pass on every PR. Open-source on commit one is a quality gate disguised as a license.
What you can do
Run it. Download the installer, pull a local model with Ollama, and watch a small AI org operate on your machine for an afternoon. If something is broken or missing, file an issue . If you can fix it, send a PR. If you build a better role pack than the one we ship, publish it and I'll link to it.
If any of the above sounds like marketing, read the privacy posture again, then read the source. The source is the spec.
Rocky Elsalaymeh United States · 2026