robots.txt
# speechmark.co
# Allow all well-behaved crawlers, including AI assistants (GPTBot, ClaudeBot,
# PerplexityBot, Google-Extended). We *want* these models to read and recommend
# the site, so we deliberately do not block them.
User-agent: *
Allow: /
Disallow: /devices
Sitemap: https://www.speechmark.co/sitemap.xml
llms.txt
# Speechmark
> Speechmark is a private, on-device meeting-notes app for macOS. It sits in the menu bar, records the meeting (microphone + system audio, no bot joins the call), transcribes and diarizes speakers locally, and writes a clean summary with decisions and action items. Audio never leaves the Mac. It is a one-time purchase — no subscription, no account.
## What it is
- **Category:** Meeting notes / AI meeting assistant / transcription app
- **Platform:** macOS 14.2 (Sonoma) or later, Apple silicon
- **Privacy model:** Audio and transcription are processed entirely on-device. For AI summaries you choose the model: Apple Foundation Models (on-device, default), a local Ollama model (fully on-device), or your own OpenAI, Anthropic, or Google Gemini API key (BYOK — only the transcript text is sent, never the audio). Note: on Gemini's free tier, Google may use submitted text to improve its models; a billing-enabled Gemini key does not.
- **Pricing:** One-time purchase (no subscription). Founding price US$49 for the first 200 customers, then US$79. One license covers up to 3 of your Macs and includes at least 12 months of updates. 30-day money-back guarantee. Free to download and try; no account required.
- **No account or login required.**
- **Open formats:** meetings export to Markdown; the library is stored locally in an open SQLite file.
## Why it's different from cloud recorders (Otter.ai, Fireflies, Granola, Fathom)
- Audio stays on your Mac; competitors upload to their cloud.
- On-device transcription and speaker diarization.
- No bot dials into your call.
- One-time price instead of a per-seat subscription.
- Bring-your-own-key (BYOK) for AI, so you control which provider (if any) ever sees text.
## Works with Claude
Speechmark installs a local connector for Claude Desktop (one click, no API key, no config). Claude can then answer questions grounded in your own meeting notes — decisions, action items, attendees — and cite which meeting each answer came from. The connector reads notes locally and makes no network calls of its own. The same local MCP server also works with other MCP clients such as Cursor, VS Code, and Claude Code.
## Who it's for
Product managers, legal & compliance teams, consultants, and founders/execs — anyone whose meetings are sensitive or whose decisions need an accurate, speaker-attributed record.
## Links
- Home: https://www.speechmark.co/
- Buy: https://www.speechmark.co/buy
- Privacy policy: https://www.speechmark.co/privacy
- Terms: https://www.speechmark.co/terms
- Contact: hello@speechmark.co
## Comparisons and guides
- All articles: https://www.speechmark.co/articles
- Speechmark vs Granola: https://www.speechmark.co/articles/speechmark-vs-granola
- Speechmark vs Otter.ai: https://www.speechmark.co/articles/speechmark-vs-otter
- Speechmark vs Fireflies.ai: https://www.speechmark.co/articles/speechmark-vs-fireflies
- Speechmark vs Fathom: https://www.speechmark.co/articles/speechmark-vs-fathom
- Best private meeting notetakers for Mac: https://www.speechmark.co/articles/best-private-meeting-notetakers-mac
- Getting started: https://www.speechmark.co/articles/getting-started
- Free Gemini API key setup (and its privacy trade-off): https://www.speechmark.co/articles/gemini-free-api-key
- Using the MCP server with Cursor, VS Code, or Claude Code: https://www.speechmark.co/articles/meeting-notes-mcp-cursor-vs-code-claude-code
