Local by design, not by policy

Meeting notes & action items.

100% on your machine. Free to try.

Clearminutes records your call, transcribes every word on-device with Whisper, labels who said what, and writes a clean summary with action items the second you stop. Audio, transcripts, and summaries never leave your machine.

Download free Try it in your browser

Your meetings never leave your device
No bot joins your call
Runs offline
macOS, Windows

Runs 100% locally by default · Cloud AI keys optional (text-only, off by default) · Telemetry off, never your meeting content · macOS, Windows · No account to try

🔒 Audio processed on-device

🚫 Never used to train AI

☁️ Cloud AI opt-in

Domain clean ✓

Clearminutes app screenshot showing transcription and summary

Clearminutes app screenshot in dark mode showing transcription and summary

How we compare

Clearminutes vs the rest.

No bots. No cloud by default. No monthly bill. See how we stack up against Otter.ai, Fireflies, and Fathom.

Feature	Clearminutes	Otter.ai	Fireflies	Fathom
Real-time transcription	Local + offline	Cloud only	Cloud only	Cloud only
Audio leaves your device	Never by default	Always	Always	Always
Meeting bot joins call	No bot	Bot required	Bot required	Bot required
Works without internet	✓	✗	✗	✗
Free tier	✓ Unlimited transcriptions	Limited minutes	Limited minutes	Limited meetings
One-time purchase option	Available	✗	✗	✗
Open-source models	✓ 20 local + Ollama	✗ Proprietary	✗ Proprietary	✗ Proprietary

Who it's for

Made for the conversations
your work depends on.

Healthcare

No PHI in the cloud — your meetings never leave the clinician's machine. Speaker labels separate clinician from patient (Pro). Runs offline on locked-down clinic networks.

How healthcare teams use it →

Legal

Privilege preserved: nothing is uploaded to a third party. Verifiable on-device processing for risk committees. PDF and outline export for exhibits and billing.

How legal teams use it →

Consulting

Client commitments captured as trackable action items. TickTick sync (Pro) and client-ready outline export. A reliable record for billable hours.

How consultants use it →

Sales

No bot announcing itself in discovery calls. Next steps captured automatically. Import recorded calls and summarise them (Pro).

How sales teams use it →

Research

Sensitive interview audio stays local. Speaker-separated transcripts for qualitative coding (Pro). Offline transcription for fieldwork.

How researchers use it →

More use cases

Guides for privacy-first, offline, no-bot, and free-tier ways of working. Find the one that fits.

Browse all use cases →

See it in action

Screenshots from the app.

Real screenshots from Clearminutes running on macOS. Dark mode and light mode supported.

Real-time transcription

Words appear live as people speak, with speaker labels applied automatically.

Clearminutes instant AI summary view showing structured summary and action items generated when a meeting ends

Instant AI summaries

The moment your meeting ends, a structured summary and action items are ready.

Clearminutes action items extracted from a meeting and synced to a TickTick inbox with one click

Action items + TickTick sync

Extracted tasks sync directly to your TickTick inbox with one click.

Clearminutes local AI settings panel for choosing Whisper model and GPU backend with on-device processing

Local AI configuration

Choose your Whisper model, GPU backend, and keep everything on-device.

Try it yourself

Transcribe your own voice, in your browser.

Record your voice right now and watch Whisper transcribe it in your browser. No download, no account, no cloud.

Browser demo

Transcribe your voice in real time. A tiny AI model (~31 MB) downloads once, then runs entirely on your device. Nothing is sent to any server.

Try it now

Privacy by architecture

Your meetings
never leave
your machine.

Most AI meeting tools send your audio to their servers. Clearminutes is built the other way around: the AI runs on your device, not ours. There's nothing to intercept because nothing is transmitted.

🎙️

Audio stays on-device

Your microphone audio is captured and processed entirely within Clearminutes on your machine. It is never streamed, uploaded, or buffered to any external server.

🤖

AI runs locally

20 built-in models (Gemma, Qwen, Llama, Mistral, Phi & more) run entirely on your device with no internet. Connect your own Ollama instance for even more local models. If you optionally enable a cloud provider (OpenAI, Anthropic, Groq, OpenRouter), only the transcript text is sent: never audio.

🗄️

Data stored locally

Transcripts, summaries, and action items are stored in a local database on your machine. You own the files. Optional encrypted cloud sync is available for multi-device access: but it's always opt-in.

🎙️ Audio on-device ⚡ Summaries on-device ✈️ Works offline 🧬 Open-source whisper.cpp 📡 Verifiable privacy monitor pending

Data flow: default configuration

🎙️ Meeting audio captured On device

↓ processed locally

📝 Real-time transcription On device

↓ 20 local models + Ollama

⚡ AI summary + action items On device

↓ written to local storage

🗄️ Stored in local database On device

✕ audio never sent to internet

☁️ External servers / cloud AI Never by default

Optional: Connect your own Ollama instance for more local models, or bring a cloud API key (OpenAI, Anthropic, Groq, OpenRouter) for cloud summarisation. Only the text transcript is sent: audio is never transmitted. Disabled by default.

How it works

From talking
to done.

Join your meeting as normal

Clearminutes runs in the background and captures your system audio directly. No bot joins the call and no announcement appears in the participant list.

Menu-bar capture
Mic + system audio
No bot in the call

Transcription happens in real time

Words appear live as people speak. Speaker labels are applied afterwards: rename them to match your attendees.

On-device Whisper
Live transcript
Speaker labels (Pro)
Works offline

Summary and actions land instantly

The moment you stop the recording, your AI summary and action items are ready. Review, edit if needed, then export or share.

20 local models + Ollama
Cloud keys optional (text-only)
Action items + TickTick sync (Pro)
PDF & outline export

Platform Engineering Weekly LIVE · 12:04

James So the Flux CD upgrade. I think we should defer that. The risk assessment isn't done and Q3 is more realistic.

Sarah Agreed. I can have the DORA metrics baseline doc done by Friday if that helps.

Tom I'll put together the observability vendor shortlist and share it with the team by Wednesday.

James Perfect. Budget for the observability tooling is confirmed on my end. Go ahead.

Clearminutes action items with TickTick sync

Pricing

Pricing.

Start free. Upgrade when you need more. No hidden fees.

Free

^£0

forever

Real-time transcription
AI summaries
Action item extraction
Meeting history
Speaker identification
Multiple devices
TickTick task sync
Export to Outline

Download free

Built for people who meet a lot.

★★★★★

"I used to spend 20 minutes after every call writing up notes. Now I just review the summary, tweak anything, and I'm done in 2. It's genuinely useful."

Alex Reeve

Engineering Manager · Fintech startup

★★★★★

"The action item extraction catches things I said I'd do that I'd already filed as 'probably going to forget'. The accountability alone is worth it."

Maya Okafor

Independent Consultant

★★★★★

"Privacy was my main concern. I can't have client calls going to some server somewhere. The local processing model was the deciding factor. That and it actually works."

Dan Whitfield

Freelance Product Designer

FAQ

Common questions.

Does Clearminutes send my audio to the cloud? +

No. And this is a core design decision, not just a setting. Your audio is captured and processed entirely on your device using an on-device AI model. It is never streamed, uploaded, or buffered externally. If you enable optional cloud sync for cross-device access, only the finished text transcript is synced (encrypted in transit and at rest). Audio is never part of that.

Can I use cloud AI like GPT-4 or Claude for better summaries? +

Yes: optionally. Clearminutes ships with 20 built-in local models (Gemma, Qwen, Llama, Mistral, Phi & more) and supports connecting your own Ollama instance for even more local options. For cloud AI, bring your own API key from OpenAI, Anthropic, Groq, or OpenRouter. When enabled, only the finished text transcript is sent to that provider. Your audio is never transmitted. This is an opt-in setting, off by default. Your API key is stored locally on your device and never shared with us.

Which meeting platforms does it work with? +

Clearminutes works with any meeting platform because it captures audio directly from your system: Zoom, Google Meet, Microsoft Teams, Webex, Slack Huddles, or a plain phone call. If your computer can hear it, Clearminutes can transcribe it.

How accurate is the transcription? +

In clear audio conditions, word-error rates are below 5% for English. Accuracy varies with audio quality, accents, and background noise. Speaker identification works best with 2 to 6 distinct voices. You can always correct the transcript before exporting.

What happens to my data if I cancel? +

Your local data is always yours and stays on your device. If you've used cloud sync, you can export all your meetings as JSON or markdown at any time. When you cancel, your account data is retained for 30 days then permanently deleted.

Can I use Clearminutes in languages other than English? +

Transcription currently supports English, Spanish, French, German, Portuguese, and Japanese. AI summaries are generated in the same language as the transcript. More languages are being added: check the roadmap for details.

Is there a team or business plan? +

Not yet: Clearminutes is currently focused on individual professionals. A Teams plan with shared meeting history, collaborative action tracking, and centralised billing is on the roadmap for later in 2026. Join the waitlist to be notified.