WhisperBrain

Product Introduction

1. Definition: WhisperBrain is a local-first, privacy-focused AI meeting assistant for macOS. It is a desktop application that automatically transcribes, analyzes, and archives video conference calls directly into a user's Obsidian vault as searchable Markdown files.
2. Core Value Proposition: It exists to solve organizational memory loss in teams by creating a persistent, private "decision ledger." Unlike cloud-based notetakers, it operates invisibly on your Mac, ensuring meeting audio and voice biometrics never leave the device, while transforming conversations into actionable, interconnected knowledge within your existing second-brain workflow.

Main Features

1. Invisible, Local Meeting Detection & Recording: WhisperBrain passively monitors system audio to detect active meetings on Zoom, Google Meet, and Microsoft Teams (both native apps and browser tabs). It does not join as a bot participant. Using macOS's Core Audio, it captures the audio stream already playing on the Mac, ensuring no other attendees are aware of the recording. A "dynamic notch" UI appears around the MacBook's camera notch to provide live transcription and a single stop control.
2. On-Device Transcription & Speaker Diarization: The app performs Automatic Speech Recognition (ASR) and speaker separation locally using downloaded models (e.g., OpenAI's Whisper or similar). It creates voiceprints for each speaker, which are stored exclusively on the local machine. This enables accurate attribution of quotes (e.g., "Priya: So we go with Postgres") without sending sensitive audio data to the cloud. It supports over 40 languages and can handle multilingual conversations within a single sentence.
3. AI-Powered Decision Ledger & Contradiction Checker: This is the core innovation. WhisperBrain uses inference (locally via on-device models or optionally via cloud APIs) to analyze transcripts. It doesn't just summarize; it extracts specific decisions, documents the reasoning behind them, and files them in a structured Decisions.md ledger within Obsidian. It can also perform semantic analysis to flag when a new decision in a meeting contradicts a previously recorded one, preventing context drift.
4. Native Obsidian Vault Integration: The application writes directly to a user-specified Obsidian vault directory. Output is pure, standard Markdown (.md files) with native Obsidian features like internal links (e.g., linking a decision to a project note or a person). This ensures data portability and longevity—notes remain fully accessible even if the app is uninstalled.

Problems Solved

1. Pain Point: "The Cost of Forgetting." Teams waste significant time and resources recreating context, re-debating past decisions, or making contradictory choices because the original rationale was lost in transient meeting notes or disparate chat logs.
2. Target Audience: Knowledge Workers and Teams using Obsidian. Primary personas include: Engineering Managers and Tech Leads tracking architectural decisions; Product Managers documenting feature rationale; Remote and hybrid teams needing a single source of truth; Consultants and agencies auditing client decision trails; and any individual using Obsidian as a Personal Knowledge Management (PKM) system who participates in frequent calls.
3. Use Cases: Sprint Retrospectives and Planning: Capturing the "why" behind task prioritization. Technical Design Reviews: Archiving the trade-offs discussed for choosing one database or API design over another. Client Check-Ins: Maintaining a verifiable record of agreements and change requests. Board/Stakeholder Meetings: Creating an auditable trail of strategic decisions and action items. Onboarding New Team Members: Providing a searchable history of project evolution beyond static documentation.

Unique Advantages

1. Differentiation vs. Competitors (Otter.ai, Fireflies.ai, Grain): Competitors are cloud-first SaaS platforms where all audio and data is uploaded and stored on their servers. WhisperBrain is local-first and privacy-by-design. It avoids the "another dashboard" problem by integrating into the user's existing tool (Obsidian), whereas competitors create proprietary, siloed note repositories. Its focus is less on real-time collaboration and more on long-term knowledge preservation and decision integrity.
2. Key Innovation: The "Decision Ledger" concept. Moving beyond transcription and summary to a structured, queryable database of decisions-with-context. The combination of local audio processing (for privacy) with optional cloud inference (for speed and advanced multilingual analysis) offers a flexible hybrid model. The contradiction detection feature proactively maintains logical consistency in organizational memory.

Frequently Asked Questions (FAQ)

1. Is WhisperBrain secure and private for confidential business meetings? Yes, WhisperBrain is designed with maximum privacy. All audio recording and speaker identification (voiceprinting) occurs locally on your Mac. Audio data never leaves your computer. Only if you explicitly enable cloud processing for summaries are transcript texts sent to external AI models; the original audio remains on-device.
2. How does WhisperBrain work without joining the meeting as a bot? The application functions at the operating system level, accessing the audio output stream of your Mac. It "listens" to the same audio you hear through your speakers or headphones. There is no virtual participant, so other attendees see no indication of its use, though users must always comply with local recording consent laws.
3. What happens to my notes if I cancel my WhisperBrain subscription? All your notes are yours permanently. They are stored as standard Markdown files in your local Obsidian vault. If you cancel a cloud subscription, you lose access to fast cloud processing and syncing, but you retain all historical data and can continue using the fully functional local, free tier indefinitely.
4. Can WhisperBrain integrate with other note-taking apps like Notion or OneNote? Not directly. WhisperBrain's architecture is built around direct Markdown file generation for Obsidian, a tool built on local Markdown files. This ensures data ownership. You can manually copy Markdown content to other apps, or use Obsidian's export functions, but there is no native API integration with cloud-based note platforms in the current release.
5. What are the system requirements and does it work on Windows? WhisperBrain is currently a macOS-only application, requiring macOS 12 (Monterey) or later. It leverages native macOS frameworks for audio capture and security. A Windows or Linux version is not available, as the core technology is built around the macOS ecosystem.