App

IFS Journal

The journaling that helps you understand yourself

web-app

IFS Journal is a journaling app for Internal Family Systems work. You write entries; Claude finds the "parts" speaking through your writing (Managers, Firefighters, Protectors, Exiles) and anchors each one to the exact passages it came from.

How it works

  • Grounded in your words. Every part the model names is backed by cited passages from your journal, never free-form inference, using Anthropic's Citations API.
  • Citations that survive edits. Highlights are anchored the way Hypothesis anchors annotations (W3C text-quote and text-position selectors with a fuzzy fallback), so editing an entry does not break them.
  • Measured, not vibes. Extraction quality is tracked against synthetic personas, each with an intended cast of parts and 40 entries, from terse logbook writers to long-form introspection. Every model or prompt change is replayed against the same frozen entries and graded.
  • Private by default. Every request is scoped to the signed-in user; nobody else can read your entries.

The source is not public.