Meeting autopilot for macOS

AnyBrief records the meeting and assembles the brief

Start recording manually, from the calendar, from a visible window or app process, or through the local API. AnyBrief records system audio and microphone, transcribes locally through FluidAudio or whisper.cpp with speaker diarization, then prepares a Markdown brief through Ollama, an API or a trusted CLI.

Download for Mac See features Source code on GitHub What's new in 1.0.14

Version 1.0.14 · signed and notarized by Apple · universal build for Apple Silicon and Intel Macs.

AnyBrief Dashboard

Autopilot is on

calendar · windows/processes · local API · CLI

15:00

Product team — weekly sync CalDAV

Next scheduled recording

AnyBrief fills in the meeting title, records system audio and microphone, shows input levels, stores calendar metadata and exposes ready artifacts to local agents.

14:59 autostart

2 audio tracks

MD + ZIP brief, transcript, archive

Two local transcription engines — one finished transcript

Choose FluidAudio speed or whisper.cpp flexibility. Both paths identify speakers, and AnyBrief never sends your audio to a cloud transcription service.

A provider for each kind of meeting

Switch providers directly in Transcription settings. FluidAudio diarization remains available for whisper.cpp results.

F FluidAudio STT NVIDIA Parakeet TDT 0.6B v3 · Core ML · 25 languages Fast

              W
              
                whisper.cpp + FluidAudio
                99 languages · word timestamps · speaker diarization
              
              Flexible

Model, language and acceleration under control

Download the Whisper model separately, choose a language or automatic detection, enable Metal, and check local technology readiness in one place.

tiny base small medium large-v3-turbo

Core MLChecked in the app

Apple Neural EngineChecked in the app

Metalwhisper.cpp acceleration

Models and CLIReadiness and cache path

What it actually looks like

Real screens from the app: autopilot, live transcript, prompts, LLM connections and settings — no mockups, no renders.

AnyBrief Status screen: active recording and recent meetings — **Status**Active recording, live system-audio and microphone levels, and recent meetings at a glance.

AnyBrief Autopilot screen: today's meeting schedule with a Join button — **Autopilot**Today's calendar, CalDAV auto-start and a one-click Join right from the app.

AnyBrief Live screen: live transcript and LLM output during the call — **Live**Live transcript and LLM processing during the call, driven by your own prompt.

AnyBrief Post-processing screen: automatic summary settings — **Post-processing**Automatic summary with prompt, LLM connection and speaker-context selection — plus dedicated tabs for transcript cleanup and export.

AnyBrief Prompts screen: reusable prompt library for summary, Live and transcript cleanup — **Prompts**Your own prompt library for every task: summary, Live notes and transcript cleanup.

AnyBrief Settings → LLM screen: connection pool of CLI, Ollama and OpenAI-compatible APIs — **Settings → LLM**A pool of LLM connections — CLI tools, Ollama and OpenAI-compatible APIs — each with its own timeout, retries and chained fallback.

AnyBrief Permissions screen: system permission status — **Permissions**A transparent view of system permissions — microphone, screen recording and notifications — no hidden requests.

The new version covers the full meeting path

AnyBrief is now a modular workstation for calls: recording, autopilots, summary providers, history, settings, logs and local integrations live in one app.

System audio and microphone recording

Records both sides of the call into separate tracks and shows live signal indicators and current source names.

⏱

Calendar, window and process autopilots

CalDAV starts recordings from the schedule, while Window Observer watches the active app or all visible windows and matches the app/process owner name plus window title.

👥

FluidAudio or whisper.cpp

Choose fast FluidAudio STT or multilingual whisper.cpp. For whisper.cpp, AnyBrief combines word timestamps with FluidAudio speaker diarization.

🎙

Resilient microphone capture

Start with the MacBook microphone, put on AirPods, and AnyBrief restarts only microphone capture while system audio keeps recording.

Briefs with meeting context

The Markdown output includes decisions, action items, recognition settings, calendar frontmatter, links, participants, warnings and metadata.

Multiple summary providers

Use local Ollama, OpenAI-compatible chat completions APIs or trusted CLI tools such as Codex, Claude and opencode, including subscription-backed workflows that avoid extra API spend.

⌁

Control during the call

Microphone pauses write silence into the mic track, status shows echo cancellation, and degraded mode preserves system audio if the microphone fails.

◌

Dashboard for repeat work

Meeting history, rename/delete actions, summary/Finder opening, settings sections, permissions, notifications, logs and updates.

↗

Local API and files

Agents can start and stop recordings, read jobs, meetings, transcripts, summaries, logs and settings through the localhost API.

✎

Reusable prompt library

Your own prompts for summary, Live notes and transcript cleanup, matched by meeting title patterns, each with its own LLM connection per task.

⇄

Pooled LLM connections with fallback

Multiple CLI, Ollama and API connections in one pool, each with its own timeout and retry count, CLI sandboxing from read-only to full access, and automatic fallback to the next connection on failure.

⤓

Rule-based summary export

Rules matched by meeting title drop finished summaries into the right folders with your own filename template — no manual copying after every call.

More than a file transcriber

AnyBrief is not about manually importing finished files. It is for live calls and the work after them: the app starts recording, watches audio, saves artifacts and exposes a local interface for automation.

What AnyBrief already does

Records call system audio and microphone audio on Mac.
Auto-starts and auto-stops from CalDAV calendar events.
Automates recording from the active app or all visible windows through Window Observer.
Matches Window Observer rules by app/process owner name and window title, with a stable match before recording starts.
Provides a local HTTP API for agents and localhost automations.
Restarts only the microphone when the input device changes, including AirPods.
Lets you choose FluidAudio STT or whisper.cpp, a Whisper model and recognition language.
Checks Core ML, Apple Neural Engine, Metal, command-line tools and local models directly in Settings.
Protects the timeline: silence in the mic track instead of compressed recordings.
Creates post-meeting summaries with frontmatter and calendar data.
Supports Ollama, OpenAI-compatible APIs and CLI summary providers, including existing CLI subscriptions instead of API-token spend.
Stores recordings, logs, transcripts and bundle.zip locally.

AnyBrief focus

Fewer manual steps before and after each call.
Clear recording status: microphone source, system audio, echo cancellation, levels and silence.
Providers, automation sources and settings live as modules, not as hidden flags.
Meetings can be renamed, opened, deleted and passed to agents.
Integrations through a local API instead of manual file copying.
Privacy and cost control: work artifacts stay on your Mac, and summaries can use local Ollama or your subscription-backed CLI tools.

Download AnyBrief for Mac

Version 1.0.14: manual recording and autopilots, FluidAudio or whisper.cpp with diarization, local API, and summaries through Ollama, APIs or subscription-backed CLI tools.

Download for Mac