Live copilot · EN/PT/ES meetings · no bot

Minuta listens to the call.
You know what to say.

The moment a question is aimed at you, Minuta catches the language and drafts a clear answer in English, Portuguese, or Spanish — grounded in your documents and everything already said. You read it and speak with confidence.

macOS · Windows · 7-day free trial

Google Meet open with Minuta's mini control listening to the meeting without adding a bot to the call.Minuta live panel over a Google Meet call, showing a detected question and a suggested copilot answer.Minuta start screen for configuring the session, language, translation, microphone, and context before listening.
No bot in the call

Minuta stays on your screen, not in the participant list

Use it with Google Meet, Zoom, or Teams without adding a notetaker to the meeting. The app listens from your desktop and stays available while the conversation happens.

You know this stuff. You just don't always show it in the moment.

You're on a client call. They ask something specific about a stack you've shipped a dozen times — and the name of the service just won't come. Or you're explaining an architecture and you know there's a clearer way to say it, but under pressure it comes out messy.

It's not a knowledge gap. It's the friction between knowing and performing in real time.

Multilingual meetings

You understand the work. Now answer in the language of the room.

Your technical judgment is already there. The hard part is the live moment: someone asks in English, Portuguese, or Spanish, the room waits, and you need the answer to come out clearly before the conversation moves on.

Minuta closes that gap with a real-time transcript, optional translation between English, Portuguese, and Spanish, and a suggested answer you can read and say with confidence in the right language.

  • Live translation across English, Portuguese, and Spanish while the other side talks
  • Didn't catch that? The transcript is right there — scroll back without breaking the call
  • Copilot drafts the answer in the meeting language, at your technical level
  • Works on Zoom, Teams, Google Meet — no bot joins, nobody sees it

Use cases

The question lands. The answer is already forming.

Real moments from technical calls in English, Portuguese, and Spanish — and what the copilot suggests you say.

Mid-call, the client asks a pointed architecture question.

They asked:

How would you handle retries and idempotency if that payment webhook fires twice?

Say this:

I'd make the handler idempotent — key each event by its provider ID and skip anything already processed. Retries become safe by design: a double-fire just hits the dedup check. I set this up on a billing integration last year.

Pro only

Your copilot inside the meeting

This is what Minuta is really about. The copilot listens to the meeting and automatically detects when a question is directed at you. One click gives you a suggested response grounded in your documents, what's already been said in the meeting, and real-time web search. Available on the Pro plan only.

  • Automatically detects questions in the live transcript
  • Suggested responses based on your documents + meeting transcript
  • Real-time web search to enrich the answer (optional)
  • Load PDFs, notes, or texts as context before starting
Copilot● REC 14:52

Live transcript

14:21Maria: Let's go over the next steps for the proposal.
14:38Carlos: The timeline you mentioned earlier — does that still hold?
Question detected
14:52João: "What makes your product different from competitor X?"
💡What to say
Suggested response

Based on the proposal loaded and what's been discussed, you can highlight: native integration with existing tools, dedicated support included in all plans, and the volume discount mentioned earlier that applies here.

Sources:📄proposal.pdf💬transcript🌐web search

Trust layer

Local capture. Cloud AI chain. No bot in the meeting.

Minuta stays a desktop app for system audio capture, then sends the minimum required meeting data through reviewed cloud paths for transcription, summaries, speaker ID, and copilot answers.

  1. 01

    Your computer captures the call

    Zoom, Teams, Google Meet, and browser calls keep running normally. Minuta listens from your machine; nothing joins as a participant.

  2. 02

    Cloud AI chain processes the audio

    Deepgram transcribes the audio (OpenAI Whisper and Groq run as automatic fallback). OpenAI drafts summaries and copilot answers (Anthropic Claude Haiku and Groq run as fallback). Modal runs pyannote for speaker detection. Payloads are narrow by design and reviewed in code.

  3. 03

    You keep the useful record

    The transcript, speaker labels, summary, and copilot answers stay locally in the desktop app — ready to review, export, or delete.

Meeting audio is captured locally, processed through a reviewed multi-provider cloud chain, and saved back to your desktop app.

How it works

Three steps. Then just show up to your meetings.

  1. 01

    Tell Minuta who you are

    Upload your resume, projects, stack, and any docs that matter. The more context, the sharper the answers. All stored in your profile.

  2. 02

    Open Minuta before the call

    Works with Zoom, Google Meet, Teams, or any platform. No bot joins the meeting. Invisible during screen share.

  3. 03

    Get answers at your level

    When a question comes up, Minuta suggests a response in your tone, with your experience, in your technical domain. Glance, adapt, answer.

Why Minuta

This isn't another AI notetaker.

Deep context

Your profile, not a generic resume read

Other copilots give you Wikipedia answers. Minuta speaks from what you know, what you've built, and how you actually explain things.

Invisible

No bots, no disruption

Nothing joins your call. The other side sees nothing. It's just for you, on your screen — even during screen share.

Technical calls

Built for people who live on technical calls

Consultants, engineers, architects, sales engineers. If your job is explaining complex things live, Minuta was designed for you.

Under the hood

Lightweight and fast, running silently.

Speed

Fast cloud processing

Audio capture stays local on your desktop, but transcription, speaker ID, and copilot run over secure, zero-retention cloud paths using fast cloud models. The heavy models are gone from your laptop.

Pipeline

Transcribes, translates, and identifies speakers

Detects voice in real time, automatically transcribes English, Portuguese, and Spanish, and translates when you need a second language on screen. When you stop, it identifies who spoke each segment — no manual steps.

Output

Automatic meeting summary with tasks

As soon as you stop recording, the AI generates a complete meeting summary: executive summary, key points, and identified action items. Ready to paste into Notion, Obsidian, or email.

Who it's for

Who's using Minuta.

Technical consultants & contractors

Discovery calls, architecture reviews, code walkthroughs. Minuta brings the context from your past projects to the right moment.

Engineering managers & tech leads

1:1s, performance reviews, planning meetings. Fewer "let me get back to you," more direct answers.

Sales engineers

Technical demos where one question can kill a deal. Minuta keeps your battlecards at your fingertips.

Lawyer or legal team? Minuta also shows up in client meetings, negotiations, and depositions — there's a dedicated page for legal work. See how legal teams use Minuta

Privacy

What's said on your call stays with you.

  • Core transcription runs securely with transient APIs
  • Your data is never used to train models
  • Delete everything with one click
  • Optional web search, translation, account, and billing features may contact external services

Plans

Every account starts with 7 days of full access.

No model download and no meeting bot. The cloud plan keeps Minuta fast to try, then lets you stay on Free or unlock unlimited live copilot workflows with Pro.

Free
060 min / month

7 days of full access. Then 60 minutes/month for light use.

  • 7 days with everything unlocked - copilot, summary, speaker ID
  • 60 minutes/month of cloud transcription and translation after the trial
  • No meeting bot joins Zoom, Teams, or Google Meet
  • Local transcript storage plus Markdown / Obsidian export
  • No credit card required
Start the 7-day trial
Founding 100: $4.50/mo
Pro
$4.50/ month

Checking Founding-100 spots...

Unlimited cloud recording, live copilot, summaries, speaker ID, and personal context. Cancel anytime.

  • Unlimited cloud transcription and recording
  • Live AI copilot inside the meeting
  • Automatic summaries - TL;DR, key points, action items
  • Automatic speaker identification on every recording
  • Personal context: load PDFs, notes, and documents
  • Real-time web search to enrich answers
  • Priority support

First 100 customers get $4.50/month for 12 months. Stripe charges in BRL — your card issuer converts at its daily rate.

Prices in USD and EUR are indicative; Stripe charges in BRL. Your card issuer converts at its daily rate. Cancel anytime, no questions asked.

FAQ

Frequently asked questions.

Free keeps working with 1 hour/month of secure transcription and translation. The in-meeting copilot, automatic summary, speaker identification, personal context, and unlimited recording become Pro-only — subscribe any time to bring them back. No charges happen automatically; we never ask for a card up front.

Ready to answer clearly in your next meeting?

Download Minuta and use it on your next meeting. Free for 7 days, no card.