Meetings · Dictation · Voice notes

Private meeting notes.
Made on your Mac.

Daisy captures meetings, dictation and voice notes, transcribes on-device, and turns what you said into useful, searchable memory on your Mac.

Download stable1.0.7.63

Meet Daisy

One quiet tool.
Three ways to speak.

Pick a mode by what you want to capture—and where you want the words to go when you are done.

01 / Meetings

Both sides. No bot.

Record your microphone and anything playing on your Mac. Daisy separates the speakers and turns the call into a finished note.

Start: Toggle or auto-start
Captures: Mic + Mac audio
Result: Library · Markdown

02 / Dictation

Speak. Keep typing.

Hold a hotkey and talk naturally. Daisy transcribes on-device and pastes the text wherever your cursor already is.

Start: Hold a hotkey
Captures: Your microphone
Result: Current app · Nothing saved

03 / Voice notes

Catch the thought.

Tap a hotkey, say what is on your mind, and keep it as a titled, searchable standalone note in your own library.

Start: Tap to start and stop
Captures: Your microphone
Result: Library · Markdown

In the 1.0.7 beta line

Prepare the meeting.
Understand the work around it.

Daisy 1.0.7.59 introduced a workload dashboard, meeting preparation and evidence-backed plan review, plus an optional local privacy layer for cloud summaries.

Summaries · 1.0.7.57 and newer

Use the ChatGPT account already on your Mac.

Connect through ChatGPT or Codex instead of adding an OpenAI API key. Daisy shows the model and provider-reported plan window, while API-key and fully local modes remain separate choices.

Set up account-backed summaries

ChatGPT accountConnected

ModelGPT-5.6

Weekly window68% remaining

Home · 1.0.7.59

See the shape of your workload.

One period control drives activity, call time, weekday load, writing statistics and AI usage across 7, 30, 90 and 365 days.

30 days8h 24m in meetings

Before and after the call

Bring a plan. Leave with evidence.

Add an agenda as text, Markdown, PDF or DOCX. After the summary, Daisy reviews every item separately with transcript evidence, timestamps and confidence.

Confirm rollout ownerCompleted · evidence at 32:18

Optional cloud protection

Reduce what a remote AI can identify.

Daisy can pseudonymize detected people, companies and contacts locally before supported cloud summaries, while secrets and payment-card numbers stay redacted.

Protected before sendingOff by default

After the conversation

A finished note, before you move on.

Daisy turns a conversation into the useful parts while the context is still fresh—not another recording you promise yourself you will replay.

Searchable transcript

Every line stays attached to the person who said it, ready to find later.

A summary with a point

The conversation becomes a short note you can understand without replaying it.

Decisions and next steps

Commitments leave the call as concrete actions, not a vague memory.

Plain Markdown

Keep the finished record in a folder you choose and use it with the tools you trust.

Q3 product review

Ready

Summary

A smaller beta, finished properly.

Keep the launch focused, make onboarding effortless, and learn from the first group before broadening scope.

Decisions

Ship the smaller scope first Invite five early users Review onboarding after week one

“Let's make the smaller scope feel finished.”You · 32:18

The archive at work

Your conversations become working context.

Connect Claude, Cursor or another MCP client to ask across the local archive. Prepare the day, gather loose ends and understand a project without reopening every meeting.

Daily briefTuesday, July 21

Good morning

Three conversations deserve your attention today.

The beta scope is ready to confirm, Lena is waiting on the launch brief, and two action items have carried over from yesterday.

10:00Beta review2 related notes · 3 open itemsPrep

13:30Lena / launchLast spoke 4 days agoPrep

16:00Design critiqueNo preparation neededReady

Across meetingsTo close

Update the launch briefFrom Q3 product review

Send the beta inviteFrom Lena / launch

Confirm onboarding ownerFrom weekly planning

Ask the archiveProject analysis

What changed in the beta plan this week?

The scope narrowed from six onboarding steps to three. The team agreed to test with five users before adding integrations.

Answer composed from 4 local sessions

These scenarios use Daisy's local MCP connection. Your AI client reads only the sessions you choose to expose.

Local core · Open edges

Local by default.
Connected when you choose.

Daisy records, transcribes and keeps the archive on your Mac. Local AI can stay inside that boundary. External AI and destinations are optional branches you connect deliberately.

Claude

ChatGPT / OpenAI

Apple Intelligence

Ollama

LM Studio

NNotion

MCPMCP clients

↗Webhooks

Your Mac

The archive starts and stays here.

Capture
Transcription
Markdown library

01 / Local

Speech becomes text on-device.

No Daisy account and no Daisy cloud archive between you and your notes.

02 / Your AI

Choose the brain for summaries.

Use a local model, Apple Intelligence, your ChatGPT account, or a provider API key. Daisy shows the boundary before anything leaves the Mac.

03 / Open edges

Move the result, not the whole archive.

Send selected notes through configured destinations, webhooks or MCP workflows.

Read the full privacy boundary ↗

Comparison

One local product instead of a stack of subscriptions.

A complete comparison across meetings, dictation, storage, automation and AI access. A check means the capability is built in; a dash means no directly comparable core workflow is documented.

Capability	Daisy	Wispr Flow	Granola	Otter	Fireflies
Meeting recording	✓	✓	✓	✓	✓
Dedicated voice-note mode	✓	Scratchpad	—	—	—
Dictation in any app	✓	✓	—	—	—
Speaker separation	✓	—	✓	✓	✓
Meeting screenshots	✓	—	—	—	—
Pre-meeting context	Local brief	—	Cloud	Cloud workflow	Cloud
Cross-meeting insights	Local	Usage only	Cloud chat	Cloud chat	Cloud analytics
CRM and destinations	Destinations + webhooks	Dictates into CRM	Native + API	Native + API*	Native + API
Languages	12 + auto-detect including Russian	100+	Multi-language	6	100+
Transcription	Local	Cloud	Cloud	Cloud	Cloud
Primary archive	Local Markdown	Cloud account	Cloud workspace	Cloud workspace	Cloud workspace
Summary model choice	Local or your provider	Wispr models	Granola models	Otter models	Fireflies models
Local protection before cloud AI	Pseudonymization	—	—	—	—
Works offline	✓	×	×	×	×
API / automation	Local tools + webhooks	—	REST API	Enterprise API*	GraphQL API
Open source	✓	×	×	×	×
MCP and agent access	Local	—	Cloud	Cloud	Cloud
Daisy/vendor account required	No	Yes	Yes	Yes	Yes
Entry price	Free during beta	$0 / Pro $15	$0 / Business $14	$0 / Pro $8.49*	$0 / Pro $10*

Checked against official product materials on August 19, 2026. * Otter API is Enterprise-only; Otter and Fireflies prices are per user/month with annual billing.

Proof without theater

Built in public.
Specific about the limits.

Trust comes from inspectable software, clear data boundaries and claims that can be checked—not an invented accuracy number or a wall of anonymous quotes.

01 / Source

Apache 2.0.

Read the code, build it yourself or inspect exactly how the data path works.

Open GitHub ↗

02 / Platform

Native to the Mac.

Signed and notarized, built around Apple's audio, security and on-device ML stack.

View the security model ↗

03 / Accuracy

No magic percentage.

Daisy uses published speech engines, re-runs speaker separation offline and admits where overlap remains hard.

Read the benchmark protocol ↗

Freeduring beta

macOS 14+Apple Silicon

No accountdownload and start

Your providerexternal AI billed directly

Before you record

The practical questions.

Privacy, compatibility, AI and what actually happens when you press Record.

What is Daisy?

Daisy is a private meeting notes, dictation and voice notes app for Mac. It records both sides of a call without a bot, transcribes on-device and keeps the result as local Markdown.

Does a bot join the call?

No. Daisy captures your microphone and the audio already playing through your Mac. Nothing joins the meeting, announces itself, or appears in the participant list.

Which meeting apps work with Daisy?

Daisy works at the Mac audio layer instead of relying on a meeting-platform integration. That means calls playing through your Mac can be captured across apps such as Zoom, Google Meet, Microsoft Teams, Telegram and Webex.

What stays local?

Recording, transcription and your Markdown archive live on your Mac. If you choose an external summary provider or destination, that specific content is sent directly from your Mac to the provider you selected — Daisy is not a proxy and does not keep a cloud archive.

Do I need an AI API key?

Not for recording or transcription. For summaries you can use a supported local option, Apple Intelligence on compatible macOS versions, connect the ChatGPT account already signed in through ChatGPT or Codex, or add your own provider API key. API keys stay in macOS Keychain; account-backed requests use your ChatGPT plan limits.

Where do the notes go?

Into the folder you choose, as plain Markdown. That can be a regular local folder, iCloud Drive, an Obsidian vault, or another place you control.

How do integrations work?

Daisy can expose the local archive to compatible AI clients through MCP and can send finished results through configured destinations or webhooks. Each connection is optional and set up explicitly by you.

Which Macs are supported?

Apple Silicon Macs running macOS 14 or newer. Some system AI features require a newer version of macOS, but recording and on-device transcription do not require an Apple Intelligence-capable Mac.

Which languages does Daisy support?

Daisy offers transcription in 12 selectable languages: English, Russian, Ukrainian, Polish, Spanish, German, French, Italian, Portuguese, Japanese, Korean and Chinese. Auto-detect is also available. Summary-language support depends on the AI provider you choose.

Do I need to tell participants that I am recording?

Recording and consent laws vary by location. You are responsible for giving any notice and obtaining any consent required where you and the participants are located.

How much does Daisy cost?

Daisy is free during beta. Any external AI usage is billed directly by the provider you choose. Final post-beta pricing will be published before the beta ends.

Made for the Mac in front of you

Try Daisy on your
next conversation.