AI-life balance
Your AI life, measured.
One command turns activity across your coding agents into a clear view of your day and night. Change one thing, then see whether it helped.
Start measuringGet the iPhone appConnect to web and mobile to see AI activity, sleep, and recovery—in one view.
no hustle leaderboardno employee monitoringyour account · your history
Your AI life report
Your week has a pattern.
One view across the coding agents you already use.
afterhoursyour human dashboard
todaythis week30 days
Jul 14—20time back · estimate+7.4h
Your agents worked for 31 hours. They needed 4h 12m of your actual attention.
↗ 2.1h more than last week4/5nights
protected
protected
your attention rhythmAgent activity vs. you checking in
agentsyou
after hours
8amnoon4pm8pmmidnight
11:47pm: agents were still productive. Your correction rate was not.
day + nightYour whole week, in one view
AI activitysleeprecovery
See your AI day and your night together.
AI saved you 7.4 hours.
You gave 5.1 of them back by launching more work after your planned stop time.
69% of your AI dividend went straight back into the laptop.
Let them run. Don't supervise.
Two agents can finish without you. Your next useful check-in is tomorrow at 9:00am.
This would create a local boundary—not stop or kill your agents.the crewWhere your AI time went
metadata only · prompts stay yoursCClaude Code19.2h2h 44m attention
⌘Codex9.4h1h 03m attention
✦Gemini2.4h25m attention
Not another productivity cop
The output shows what AI produced. Teeras shows what it asked of your time and attention.
Local content boundary
Read logs locally, keep counts and timestamps, and discard prompts, source code, outputs, projects, and paths before sync.
Life over usage
A billion tokens is not a win. An evening returned, a walk taken, or a task finished without supervision might be.
One experiment
No guilt spiral. Try two agents instead of five, protect one stop time, and see whether the work actually changes.
The short answers
AI-life balance, without the surveillance energy.
What is Teeras?
Teeras is an AI-life balance app for people who work with coding agents. Its Afterhours report shows when Codex, Claude Code, Gemini CLI, or Cursor were active on your Mac, and the iPhone app puts that next to your sleep from Apple Health—so you can see whether late-AI nights are associated with falling asleep later, compared against your own median—then move your evening line and see whether it helped. It never reads prompts or code.
What is AI-life balance?
AI-life balance measures whether coding agents reduce human attention and protect time away from work—not merely how many tokens or tasks they consume.
Which coding agents does Afterhours track?
Afterhours currently reads counts from the Codex app and CLI, Claude Code, and Gemini CLI. Claude Code CLI history connects automatically; the Claude Code desktop app keeps its own history, so setup installs opt-in lifecycle hooks that count desktop sessions too. It detects Cursor and other installed coding agents while adapters mature. Cloud sessions that never touch your computer—like Claude Code on the web—aren't counted.
Does Afterhours upload my prompts or code?
No. Your first report is created on your computer. If you connect Teeras, activity counts and timestamps sync for your history; prompt text, source code, model output, project names, and file paths stay on your computer.
What is AI burnout?
Developers publicly describe late nights that feel voluntary, reviewing AI output instead of building, and weeks where work goes mechanical. Afterhours measures the part you can measure — whether your AI day extends into your night — and leaves the diagnosis to you. More at teeras.com/ai-burnout.
How do I install Afterhours?
Run npx @teeras/afterhours setup—yourself or pasted to your coding agent. It connects your account and shows when your coding agents worked, when you checked in, and whether those check-ins reached into the night—including activity from the Codex and Claude Code desktop apps as well as their CLIs, with AI activity, sleep, and recovery together on web and mobile.
Start measuring.
See when your coding agents worked, when you checked in, and whether those check-ins reached into the night—including the Codex and Claude Code desktop apps and CLIs.
npx @teeras/afterhours setupRun it yourself, or paste it to your coding agent — it handles the rest.