Psych LM

Private, self-guided AI coaching for reflection, planning, journaling, goals, and personal context. Built around local storage, structured memory, and on-device use.

The architecture behind Psych LM is documented publicly.

The report, The Model Is Not the Product: A Dual-Pillar Architecture for Local-First Psychological Coaching, describes the app's local-first runtime, structured memory corpus, deterministic orchestration, and reliability-focused evaluation approach.

Open DOI

Local context, user control, clear limits.

Psych LM is designed to keep app-created content on your device unless you choose to export, share, back up, or contact support.

Local app data

Chats, profile context, goals, and memory records are intended to live in local app storage.

Readable exports

Exports may contain sensitive personal information and should be stored carefully.

No crisis use

Psych LM is not emergency care, medical advice, diagnosis, therapy, or crisis support.

One page for App Store review and user help.

The support page includes contact guidance, App Store links, export cautions, troubleshooting steps, and safety boundaries.

Open support page
App Store listing and support contact Local data and export questions Bug reports and device details Safety and scope limits

Local context with durable structure.

Psych LM is built around persistent memory, local processing, and a user-controlled record of what matters over time.

Persistent context

Important goals, plans, preferences, and patterns can carry forward across sessions.

Structured memory

The app is designed around durable records instead of disposable chat-only history.

User-controlled records

Local data and exports remain something the user can inspect, manage, and handle carefully.

Focused on the local-first coaching experience.

Psych LM centers on private self-guided coaching, persistent local context, memory review, profile settings, and exportable records.

Conversation continuity

Sessions can build on goals, profile context, and memory records instead of starting from zero.

Local-first records

Chats, memory, plans, and exports are designed around data the user can inspect and handle.

Practical controls

Settings, profile context, and export tools keep the experience understandable and manageable.

Built by an independent developer.

Psych LM is created by Alex Mihalcea as a local-first AI app for private reflection, planning, journaling, goals, and personal context.

Private on-device AI tools User-controlled local data Clear safety and support boundaries Practical coaching-style workflows