Anarlog is an open-source, local-first AI meeting notepad and private audio transcription application built for macOS, Windows, and Linux. Unlike conventional cloud notetakers (such as Otter.ai or Fireflies) that force intrusive recording bots into video calls and store recordings on remote servers, Anarlog operates completely bot-free by capturing system and microphone audio directly from the desktop. Built on a Tauri/Rust architecture, it gives users complete data sovereignty by supporting local on-device transcription models (like Whisper), Bring-Your-Own-Key (BYOK) APIs, local CLI/MCP tools, and end-to-end encryption.
Use Cases
Bot-Free & Confidential Meeting Transcription
Record and transcribe sensitive internal, executive, or client video calls without an awkward third-party recording bot entering the room or appearing on screen shares.
Local-First Knowledge Capture & Synthesis
Keep all transcripts, audio files, and AI-generated action-item summaries stored locally on your device to comply with strict data privacy and NDA requirements.
System-Wide Voice Dictation
Use built-in desktop dictation to transform spoken stream-of-consciousness thoughts into polished, structured written text directly inside any app.
Agentic Meeting Memory via Model Context Protocol (MCP)
Expose local meeting records, transcripts, and notes directly to developer AI assistants (like Cursor, Claude Desktop, or local agents) using Anarlog’s native local API and MCP server.
Collaborative Enterprise Team Workspaces
Share curated meeting memos, templates, and summaries securely across teams with optional end-to-end encrypted (E2EE) cloud sync, SSO, and SCIM provisioning.
Features & Benefits
Invisible Bot-Free Desktop Capture
Captures internal computer audio and microphone input natively at the OS layer, remaining completely invisible to other meeting participants.
Triple AI Model Strategy (Local / BYOK / Managed)
Flexibility to run transcription and summarization entirely on-device (offline), plug in personal LLM/STT API keys, or use hosted cloud models.
Native Model Context Protocol (MCP) & Local API
Built-in MCP server, CLI interface, and webhooks enabling external agents and local tools to query meeting history programmatically.
Open-Source Transparent Core (Tauri & Rust)
Full transparency into audio pipeline and data handling with an inspectable open-source codebase.
Structured Memo & Real-Time Note Editor
Combines live manual note-taking with automated AI summaries, speaker identification, and customizable meeting memo formats.
Zero Meeting Intrusion
Eliminates the friction, permissions, and awkwardness of having an external AI bot join Zoom, Google Meet, or Microsoft Teams calls.
Absolute Privacy & Offline Capability
Free tier allows 100% offline transcription and note processing on your machine without data ever touching a third-party server.
Developer & Agent-Ready
Local API, CLI, and native MCP support turn meeting transcripts into queryable contextual memory for agentic workflows.
Cons
On-Device Compute Overhead
Running local speech-to-text (STT) and LLM models offline requires decent local hardware (such as Apple Silicon or dedicated GPUs) to maintain fast turnaround times.
Manual Audio Routing on Linux/Windows
While macOS setup is seamless, certain custom virtual audio routing setups on Windows or Linux may require initial configuration.
