User Guide

Use AlmRecorder

Record or import a conversation, review who said what, and find the right moment later without sending your library to the cloud.

3 min readReviewed 2026-07-29

The useful loop

AlmRecorder is built around one repeatable workflow: capture a conversation, verify what happened, find it later, and decide whether an agent may use it.

An AlmRecorder library with recordings in English, Swedish, and Spanish.
One local memoryRecorded meetings, imports, and Voice Memos live in one multilingual library.

1. Capture or bring in audio

Record microphone and system audio as separate tracks, import common audio files, or connect Apple Voice Memos. AlmRecorder detects the language during transcription.

Record and import →

2. Make the transcript trustworthy

Every line stays linked to its speaker, timestamp, and source audio. Correct names, review uncertain cleanup suggestions, and use tags to mark projects, decisions, or follow-up context.

An AlmRecorder recording with summary, tags, named speakers, timestamps, and an MCP privacy control.
Who said what—and whenNames, timestamps, tags, and privacy stay attached to the conversation.

Review and search →

3. Ask what you remember

Use exact words when you have them, or describe the idea. Best search combines keywords and local semantic matching. Moment matches open at the relevant timestamp; a title-only match opens the recording.

4. Connect an agent deliberately

AlmRecorder includes its MCP bridge. There is no separate server package to install. Metadata, content, and writes are separate grants; recording and privacy-tag denials override them all.

Connect an AI assistant → · Understand privacy →