Skip to main content
PromptQuorumBuilt for humans. Structured for AI.
Home/Power Local LLM
Voice, Speech & Multimodal

Anarlog Review: Open-Source AI Meeting Notes, Formerly Hyprnote

·7 min readBy Hans Kuepper · Founder of PromptQuorum · Discovery engine for open-weight & open-source AI

Anarlog (version 1.4.31, released on 20 October 2026; repository last updated on 24 October 2026) is an open-source notepad that records meetings on your device and turns them into editable AI summaries. Fully local use is possible but needs an Apple Silicon Mac; every statement here comes from the project's own website, documentation and repository.

Key Takeaways

  • Maintenance: seven stable desktop releases appeared between 20 September and 20 October 2026, so no stale-project warning applies.
  • License: the community app is MIT; the enterprise folder is source-visible but commercially licensed.
  • AI choice: transcription and summaries are separate settings, each on-device, your own key or Anarlog Cloud.
  • Cost: a free plan covers on-device and own-key use; hosted AI is a paid plan.
  • Maturity: macOS is the main platform, while the Windows, Linux and mobile apps are labelled beta.

📍 In One Sentence

Anarlog is an MIT-licensed meeting notepad that captures your computer's audio, transcribes it with the model you pick and writes an editable summary.

💬 In Plain Terms

You jot notes while a call runs, the app listens on your own machine instead of joining the call, and afterwards it turns your notes and the transcript into a summary.

📌Note: PromptQuorum has not installed or run the app, so transcript and summary quality are not assessed.

Read in:
🇺🇸en🇩🇪de🇫🇷fr🇯🇵ja🇨🇳zh🇪🇸es🇧🇷pt🇸🇦ar🇰🇷ko

Anarlog is an open-source AI meeting notepad from Fastrepl, Inc. for macOS, Windows and Linux that records your computer's audio without sending a bot into the call. This review is based on the official website, documentation, privacy policy and GitHub repository, checked on 24 October 2026; PromptQuorum has not tested the app hands-on.

Anarlog Review: Open-Source AI Meeting Notes, Formerly Hyprnote

What Is Anarlog?

Anarlog is an open-source AI notepad for meetings: you type notes during a call, it records your computer's audio without joining as a bot, and afterwards it produces an editable summary. It is made by Fastrepl, Inc., and the repository describes it as an open-source alternative to Granola.

The project started as Hyprnote and briefly used the name Char; the README says the team later split the work, so char.com is now a separate productivity app and Anarlog is this meeting notepad. Some file paths in the documentation still say "hyprnote".

Get It

Anarlog is a free download for macOS, Windows and Linux; the macOS and Windows installers need no terminal.

Option
How to get it
macOSDMG for Apple Silicon and Intel from the download page or GitHub Releases
Windowsx64 installer (beta) or the Microsoft Store listing
LinuxAppImage and DEB for x64 and ARM64 (beta), an APT repository and an Arch PKGBUILD
PricingFree plan; Pro (Cloud) is $14 per month or $140 per year, in USD per the documentation
Mobile betaTestFlight for iOS and Google Play beta for Android
Websiteanarlog.so
Privacy policyanarlog.so/privacy
Source codegithub.com/fastrepl/anarlog

The pricing page renders with JavaScript and could not be read here, so the prices come from the official documentation; plan contents beyond the free and Pro split were not checked.

Requirements

Anarlog needs macOS 15 or later, and its built-in local transcription needs an Apple Silicon Mac.

macOS

What the sources say:
macOS 15 or later, with Apple Silicon and Intel builds

Windows and Linux

What the sources say:
Both are in beta: Windows x64, Linux x64 and ARM64

On-device transcription

What the sources say:
Apple Silicon only: Soniqo models, Apple Speech on macOS 26, or a whisper.cpp .bin file you supply

Local summaries

What the sources say:
A model served by Ollama, LM Studio or Unsloth, or experimental text-only Apple Intelligence on an eligible Mac

Hardware

What the sources say:
No RAM or GPU requirement is stated; speed depends on the computer and the model

Microphone detection, auto-stop and meeting-chat capture work on macOS and Linux only, and Apple Calendar integration is macOS only.

Features Confirmed by the Sources

Every item below comes from the documentation or the README; none has been tested by PromptQuorum.

  • Automatic capture. Calendar-backed meetings can start on schedule, and the app can prompt you when it detects an unscheduled call.
  • Editable notes and transcripts. You write your own memo during the call, edit transcript text, assign speakers and resume listening to add audio to the same note.
  • Summaries and chat. Templates and a customizable Auto summary produce the notes, and chat can answer questions about one meeting or several.
  • Imports. Upload an audio, SRT or VTT file, or bring in meeting history from other assistants, including Granola exports.
  • Agents and automation. A command-line tool and a local MCP server read the local database, and webhooks and starter workflows can export Markdown or file notes.

Data Flow and Privacy

Recording happens on your device, and the project says every cloud feature is optional. According to the privacy policy, last updated on 21 October 2026, notes, transcripts and meeting data are stored locally, and the README names a local SQLite database.

Where content goes depends on your settings: hosted transcription sends audio to speech-to-text providers that the policy says delete it once the job completes, hosted summaries send text through an AI gateway, and your own key or local server receives it directly. Per the policy, Cloud Sync is end-to-end encrypted, a shared note is stored readable on the servers, and the opt-in Cloud API keeps a separate readable copy.

Usage analytics (PostHog) and a sampled masked session replay run while "Share usage data" is on and can be switched off under Settings → Privacy.

📌Note: These are the project's declarations, not audit results, and PromptQuorum has not inspected the app's network traffic. The policy says crash reports stay on when usage sharing is off, while the documentation describes a separate switch for them.

Who Should Use It

  • Mac users who want transcript and summary to stay on their own machine.
  • People replacing a bot-based notetaker who prefer capture from their own device.
  • Users who already pay for an AI subscription and want to reuse it for summaries.

What We Could Not Verify

  • Transcript and summary quality. PromptQuorum did not run a meeting through the app, so accuracy, speaker labels and summary quality are not assessed.
  • Local performance. Speed and battery use of on-device models on long calls are untested; the documentation says results depend on the computer.
  • Security and privacy claims. Encryption, deletion and no-training statements are the project's own and have not been audited here.
  • Current pricing. The Pro price comes from the documentation FAQ because the pricing page could not be read.
  • Windows, Linux and mobile. The project labels these builds beta, and they were not examined.

Competitors and Alternatives

Meetily

Platforms:
Windows, macOS
Price / license:
Free Community Edition / MIT
Key difference:
Self-hosted notetaker whose summaries run through a local Ollama model; per the project, no cloud is required
Articles about Meetily (1)

Also mentioned in:

MacWhisper

Platforms:
macOS
Price / license:
Freemium / proprietary
Key difference:
One-time-purchase Mac app for local Whisper and Parakeet transcription, per the directory entry
Articles about MacWhisper (2)

Also mentioned in:

Handy

Platforms:
macOS, Windows, Linux
Price / license:
Free / MIT
Key difference:
Push-to-talk dictation into any app rather than a meeting notepad
Articles about Handy (2)

Also mentioned in:

+5 more not shown

Confirm each app's current price, license and platforms on its own listing before choosing.

Compare in directory

Frequently Asked Questions

Do I have to tell people I am recording?

The privacy policy and the documentation both say you are responsible for notice and consent where the law requires it, because recordings can include other people's voices. Anarlog does not add a participant to the call, so nobody is told automatically.

How long does Anarlog keep the audio?

Under Settings → Meetings → Audio file retention you can choose Don't save, 1 day, 3 days, 1 week, 1 month or Forever. After that period the recording is gone, while the note and transcript remain.

Which languages does it transcribe?

The Cloud picker covers a long list that includes Arabic, Japanese, Korean and Chinese. Soniqo transcribes 25 European languages, Apple Speech handles one supported language at a time, and your own provider sets its own list.

Can I use my ChatGPT or Claude subscription for summaries?

Yes, per the documentation: under Settings → Intelligence you can connect Claude, ChatGPT, Grok, GitHub Copilot or Kimi Code, so no separate API key is needed. Each device is connected separately.

Does the project train AI models on my meetings?

The privacy policy says it never uses notes, transcripts, audio or calendar data to train AI models. With your own provider, that provider's retention and training policy applies instead.

Verdict

Anarlog is an actively released, open-source way to take meeting notes without a bot, with a real choice between local and cloud AI for each step. Against that, built-in local transcription is limited to Apple Silicon, the Windows, Linux and mobile apps are beta, hosted AI is a paid plan, and none of the quality or privacy claims could be checked here. It suits Mac users who want to keep meetings on their own machine and can test the free plan first; teams with strict confidentiality needs should read the privacy policy and the provider settings carefully before relying on it.

Sources

  • Anarlog website — product positioning, download and pricing pages, checked 24 October 2026.
  • Anarlog privacy policy — data collection, providers and opt-outs, last updated 21 October 2026.
  • Anarlog documentation — installation, offline use, models and providers, data and privacy, FAQ, checked 24 October 2026.
  • Anarlog on GitHub — README, MIT license and Releases page (desktop 1.4.31), checked 24 October 2026.

← Back to Power Local LLM

MCP