Never lose a long-running job

The deploy finishes forty minutes after you stopped watching, and nobody follows up. With Paigy, the follow-up belongs to the job itself: the agent sets when to come back, and Paigy wakes it then.

How it works

Every job is a Goal: a plain statement of what done looks like, owned by the agent doing it. The agent is on it from its first write and reports its progress with contact as it works; an update you don't need yet is kept on the Goal rather than sent to you. The job sits on your phone the whole time, marked when it's waiting on you.

When the agent needs to come back later, to check the deploy, look at the test run, or follow up in the morning, it defers its Goal to that time with manage_goals. The Goal waits until then. When the time comes, Paigy wakes the agent for it.

When there's something for you, the agent reaches you on that Goal with contact: a card when it can wait, a phone call when it's blocked on you. Your answer lands back on the same Goal, so the agent picks up exactly where it stopped.

Nothing to remember

You don't track the job and you don't set reminders. The Goal carries the promise and Paigy keeps the clock. When it's done, the agent marks it done, and it leaves your list.

What this replaces

• Terminal windows kept open "just to see it finish".
• Checking a CI tab every five minutes from your phone's browser.
• Jobs that succeeded at 6:40pm and were discovered at 9 the next morning.

Works with: Claude Code, OpenAI Codex, Google Antigravity, and any MCP-compatible client — Cline, Continue, Zed, Goose, and local-model setups included.