Methodology

How IFS App works

Every session adapts to you. The onboarding itself is therapeutic. Skippable everywhere, resumable anytime, and never mining for trauma.

Adaptive 10-Stage Onboarding

The app detects pacing signals and adapts. It never asks about trauma. Every stage is skippable.

1

Intent & Goals

A single open question: what brings you here? No checkboxes, no pressure. You set the direction.

2

Emotional Baseline

Four short Likert scales: anxiety, stress, sense of control, and avoidance. Establishes a starting point.

3

Parts Identification

12 archetype cards to explore. The Critic, Perfectionist, Pleaser, Planner, and more. Add your own.

4

Behavioural Patterns

Gentle pattern questions. No trauma mining. Observable habits and reactions only.

5

Light Life Timeline

Decade-level life context. You choose how much to share. Nothing is required.

6

Protector Resistance Detection

Passive observation only. The app notices hesitations and adjusts pacing. It never asks about them.

7

Optional Deeper Layer

Only offered if readiness signals are present. Completely skippable. Always optional.

8

Trigger Mapping

Which situations activate which parts? A brief, curiosity-led mapping exercise.

9

Voice or Journaling Input

Speak or write. Audio is transcribed via Whisper. Tone seeds your personalisation profile.

10

Parts Map Preview

Your initial parts map is generated. Self at the centre, your parts in their rings. The work begins.

Architecture

Five AI Engines

Built to be technically credible, not just commercially sufficient. Safety is not optional.

Safety Engine

Crisis-pattern classifier on every input. Behavioural watcher every 30 seconds during sessions. Output post-filter with safe fallback responses.

Session Intelligence

The 6 Fs state machine: Find, Focus, Flesh out, Feel toward, Befriend, Fear (Anderson et al., 2017). Exile permission gate active throughout.

Parts Inference

Heuristic extraction of parts, intensity, Self-state, and blending signals from journal or voice. Surfaces as confirmable proposal cards — never auto-merged.

Personalisation

Adapts tone, pacing, and greetings based on your profile, recent Self-state readings, and check-in history.

AI Proxy

Abstraction layer over OpenRouter via Supabase Edge Function. GPT-4o for deep sessions, Claude Haiku 4.5 for fast inference, GPT-4o-mini for efficiency. Pluggable per task.

Session Intelligence Engine implements the 6 Fs framework from Anderson, F. G., Sweezy, M., & Schwartz, R. C. (2017). Internal Family Systems Skills Training Manual. PESI Publishing.

Ready to meet your parts?

Join the waitlist and be first to know when IFS App launches.

Join the waitlist