Local, private notes & dictation
Take meeting notes without a bot, dictate into any app, and integrate your favorite AI tools.
Download for WindowsLaunch planning sync
Release notes are drafted and waiting on final screenshots; support docs shipped yesterday.
The pricing page still shows the old tiers. Fix owned, due today.
No blockers against the Friday launch date; next check-in Thursday morning.
Launch planning sync
Release notes are drafted and waiting on final screenshots; support docs shipped yesterday.
The pricing page still shows the old tiers. Fix owned, due today.
No blockers against the Friday launch date; next check-in Thursday morning.
Meeting notes without a bot
Nothing joins your call
embral records from your computer, so other people never see an extra attendee.
It hears what your computer hears
Zoom, Teams, Meet, or a phone on speaker. Headphones are fine.
Speakers told apart as they talk
Named from your notes, on your machine.
Notes ready when you stop
Transcript, summary, and a title, written locally.
Dictation
Thanks for the walkthrough yesterday. I'd like to move forward with the pilot. Can you send over the onboarding checklist and a rough timeline for next week?
Heads up: the staging deploy is done and the migration ran clean. I left notes on the rollout doc for anything that looked off.
Idea for the launch post: lead with the privacy story, then the live transcript screenshot, then pricing. Keep it under four hundred words.
Dictate into any app
One hotkey, anywhere
Tap to talk, or hold it like a walkie-talkie.
Lands where your cursor is
Outlook, Slack, your editor: whatever has focus.
Cleaned up before it lands
Punctuation, false starts, the word "um".
Every dictation kept
Raw and cleaned, searchable later.
MCP
Give your AI assistants read and search access to your meeting notes
Claude Desktop
Registered
Claude Code
Installed
Codex
Installed
Other clients
Connect any client by hand
Your notes, in your AI tools
A built-in MCP server
Claude Desktop, Claude Code, and Codex connect in one click.
Works with the app closed
Your AI tools can read and search your meetings any time.
Search by meaning, not just words
Indexed on your machine.
Plain files underneath
Markdown and SQLite in a folder you pick.
Your audio never leaves your machine.
No account. The speech models download once, and after that it runs without a network.
Cloud transcription is optional and metered, and we store none of the audio.
Common questions
Does my audio leave my machine?
Not unless you sign in and turn on cloud processing. By default transcription and note writing happen on your own computer, and so does search. If you use embral cloud, audio streams through our servers to be transcribed and comes straight back. We store none of it.
Does embral join my calls as a bot?
No. Nothing joins the meeting and nobody sees an extra attendee. embral records from your microphone and your computer's audio, which is why it works even with headphones in.
Does it work offline?
Yes. The models download once, on first run. After that, recording, transcription, notes, and search work with no connection at all. Cloud processing is the only part that needs the internet, and it is optional.
What does embral cloud add, and what does it cost?
Cloud transcription is faster than the on-device models and labels up to 15 speakers live, and summaries come from a larger model. Plans start at $7 a month for 10 hours, with 20 hours for $9 and 40 for $13. Hours are shared between meetings and dictation, and a first subscription has a 14 day refund window.
Free, no account required.