Therapeutical AI

Should you keep working with AI for another two hours?
AI work can become inefficient fast as mental strain rises. Therapeutical AI measures both to help you save tokens, stay productive, and work sustainably.

Free research preview · Local tracking without cloud participation · No user API key

How it works

Technical signals show whether the work is paying off. Work-State questions show whether it remains clear and sustainable.

Therapeutical AI combines token use, discarded work, reprompts, and review activity with four brief AI Work-State questions.

For seven active workdays, check-ins appear every two active hours, up to three times per day. After that, they appear once per active day.

Predictions start after the first complete check-in. During the first seven active workdays they are marked as early estimates. A personalized cloud prediction can begin on the first day as soon as both worthwhile and not-worthwhile outcomes have been observed.

What is the probability that continuing AI-assisted work for the next two hours will not be worth it?

The answer is one recommendation: continue, clarify/reset, take a short break, or stop for now.

Getting started

1. Review setup. AI-tool connection and token tracking are selected by default, but no tool configuration changes until you choose the tools and confirm.
2. Complete the one-time BAT-12 baseline when offered.
3. Work normally and answer Work-State check-ins when they appear.

Code, prompts, and file names are never uploaded. Mental-strain estimates are experimental, not medical. Full details are in the bundled Privacy Policy, Terms, and license.