Welcome

Work is moving into agents. It needs a home you own.

A year ago most of us had one agent in one terminal. Now a team runs dozens, and a lot of the work starts as a message to one of them. More gets done. It also got harder to see what everyone else is working on, and to keep track of your own.

The way teams work is adjusting, and the tools with it: agent apps are growing team features, and team tools are growing agents. What a team is for hasn't changed, though. People form the mission, and the team and its agents exist to carry it out. In effect you are building one organism to do that, and we think it deserves a home of its own, with four things in it:

  • Your focus. What needs you today, and nothing else. An agent asks you by name only when a decision is yours, and shared todo lists and calendars sit next to your tasks. What a calendar and a sidebar should be when agents do most of the work is something we are still working out.
  • Your agents. Claude, Codex, Grok, and more in one window, on the subscriptions you already have. Tasks keep running when you close the app, and you can pick them up from your phone.
  • Alignment with your team. A teammate's task appears in your sidebar when it starts, and what agents make is posted to a feed. You know where the mission stands without calling a meeting.
  • The team's always-on bots. A task ends when the work is done. A bot keeps going and can own a whole area the way a person does: the release cycle, CI, research into where to distribute next. Everyone on the team asks the same one, and each bot gets only the keys its job needs.

The Happy window: Activity, Artifacts and the Launch checklist at the top of the sidebar, three bots, and tasks with their owners' faces and subtasks. A task is open beside them, where a teammate and the agent go back and forth and the agent posts a preview.

Happy is that home, built as a zero-dependency, open productivity stack. By that we mean a team should be able to do its work without stitching five other products together, and without depending on us. It is free and MIT licensed, it runs on your own computers and the ones your team connects, and the models run on the subscriptions you already pay for. You can read all of it, fork it, and host it yourself. How far along that is, and what comes next, is in Vision.

Three things are in the app. A repository is where your code is. A task is a piece of work you hand to an agent: one conversation, with a name and an owner. A bot is an agent that keeps a standing job. Quick Start walks through all three.

We use it to build it. Happy is built by two people. Between 10 October and 14 November 2026 the two of us opened 419 tasks, and our agents worked in 491 separate checkouts across 16 repositories. The examples in these docs are ours.

Where to go next

What does Happy cost?

Happy is free and MIT licensed, and the models run on your own Claude, Codex, and Grok subscriptions. Voice in the phone app is the only thing we charge for: 20 free minutes per 30 days, then up to 5 hours per 30 days on the in-app subscription. With your own ElevenLabs agent there is no limit from Happy. See Voice.