Private capture and recall

In development

Capture life. Understand it. Live better.

Oatnote turns user-initiated recordings into source-linked transcripts, notes, and proposed actions while keeping review and privacy in the loop.

Platform
iPhone · Mac
Stack
SwiftUI · SwiftData
Model
Local-first
Stage
Private development
Oatnote app iconOatnoteSHIFT KEY SOFTWARE

The product

A private, local-first system for remembering what was said and deciding what should happen next.

I am shaping the product concept, privacy model, recording and transcription pipeline, source-linked note generation, Action Inbox, and the iPhone-to-Mac experience. Oatnote treats AI output as a proposal to review—not an excuse to invent context or silently change someone’s data.

Capture deliberately

Recording starts only through an explicit user action and remains independent from network availability.

  • User-initiated recording
  • Offline-safe segment persistence
  • Timeline and source timestamps

Keep the source attached

Generated material stays traceable to the transcript instead of becoming detached AI prose.

  • Source-linked summaries
  • Speaker-aware transcript review
  • Protected local audio and timeline context

Review before writing

Tasks, reminders, and other actions are proposals until the user confirms them.

  • Action Inbox
  • Explicit approve, edit, or reject controls
  • Privacy-gated cloud processing when enabled

Product principles

AI assistance without silent authority.

Oatnote is designed around explicit capture, data minimization, source traceability, and reversible user decisions. Sensitive transcript or audio content is not treated as ordinary telemetry.

  • No hidden or covert recording
  • No raw audio or transcript upload by default
  • No invented speaker identity
  • No reminder or task creation without review
  • Deletion and export controls remain part of the product contract

Selected work

Explore the rest of the portfolio.

View all projects