Privacy and Metrics

What AIForj collects and what it never collects

AIForj is built around a local-first idea: your sessions, mood shifts, and guided-tool history should help you without becoming a pile of sensitive content sent to a remote analytics vendor by default.

This page explains the plain-English version of that promise. If you want the broader product-safety framing, read How AIForj stays safe.

Privacy guide integrity

Author

AIForj Team

Clinical review

Licensed Healthcare Provider

Last reviewed

April 16, 2026

Built for emotional first aid, not diagnosis or crisis care. Read the editorial policy to see how AIForj writes, reviews, and updates content.

Anonymous metrics preference

Default is off. If you turn this on, AIForj can send anonymous counters about tool starts, completions, and mood-shift buckets to help improve what actually works.

This preference is stored on this device. If you clear browser storage or switch devices, you will need to choose again.

What stays on your device by default

  • your mood ratings and mood-shift history
  • session history used for “For You, right now” recommendations
  • written responses you enter inside techniques and interventions
  • garden progress data and local streaks
  • your anonymous-metrics preference itself

What can be shared only if you opt in

If you turn on anonymous metrics, AIForj can send a small set of whitelisted counters to a first-party AIForj endpoint:

  • whether a tool was started
  • whether a tool was completed
  • a duration bucket such as “2 to 5 minutes”
  • a mood-shift bucket such as “up 1 to 2” when a measured intervention has both ratings
  • a short rotating anonymous client id that refreshes regularly

The goal is to understand which tools get used and which kinds of shifts are happening, without sending session content or building identity profiles.

What AIForj never sends as anonymous metrics

  • free-text responses from techniques or interventions
  • audio, voice transcripts, or message bodies
  • names, phone numbers, email addresses, or gift-note text
  • provider-search details or anything meant for crisis support
  • full raw mood histories as a default background feed

Why this is structured so narrowly

Mental-health-adjacent products lose trust quickly when they treat vulnerable moments like ad-tech fuel. AIForj is trying to do the opposite: keep emotional first aid useful, fast, and measurable without turning the underlying session content into a broad analytics stream.

If the product can learn enough from counts and buckets, it should not demand more than that.