Daisy
Daisy records meetings, dictates and keeps voice notes on your Mac — transcribed on-device, saved as Markdown you own.
For Mac users whose call audio legally or contractually cannot leave the laptop, Daisy is the most complete answer available: on-device transcription, system-audio capture without a bot, a Markdown archive you own, Apache 2.0 source, and an MCP bridge so Claude Desktop or Cursor can read that archive. The 1.0.8.32 line adds two-track export that no longer drifts and GPT-6.1 Sol as the cheaper default summarizer. Everyone else should weigh it honestly: no team workspaces, no web app, no Windows, and a cadence that ships builds every day or two. Choose Daisy for custody; choose Granola or Otter for collaboration.
Verified 1d ago · liveness 73/100 · cite: rightaichoice.com/tools/daisy
- Professionals who cannot let client or patient audio reach a cloud vendor
- Journalists, researchers and lawyers recording confidential interviews on a Mac
- Mac users who want offline push-to-talk dictation that pastes into any app
- Knowledge workers running a Markdown or Obsidian vault who want transcripts in the same folders
- Windows or Linux users — Daisy is Mac-native on Apple's audio and ML stack
- Teams needing shared workspaces, role-based access or collaborative transcript editing
- Anyone who needs a web interface or cross-device sync for the archive itself
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Daisy if your team needs shared workspaces, admin controls or transcript editing for several people, or if anyone on it uses Windows or Linux — Daisy is a single-user macOS app.
Cloud summaries bill against your own Anthropic, OpenAI, Kimi or Gemini key, and token costs follow the provider price lists Daisy mirrors — including prompt-cache writes counted at their own price
Daisy runs on the ChatGPT, Anthropic, OpenAI, Kimi or Gemini account you already pay for, or on local Gemma 4 through Ollama and LM Studio at no model cost — so the marginal cost of a meeting is tokens, not a per-seat subscription. Compared with Otter or Fireflies, which bundle cloud storage and seats into a team plan, Daisy's spend scales with how much you summarize rather than how many people you add.
In short
Daisy — Daisy records meetings, dictates and keeps voice notes on your Mac — transcribed on-device, saved as Markdown you own. Best for Professionals who cannot let client or patient audio reach a cloud vendor, Journalists, researchers and lawyers recording confidential interviews on a Mac, Mac users who want offline push-to-talk dictation that pastes into any app. Free to use.
What's new in Daisy
Checked yesterdayAcross the latest 5 updates: 5 feature updates.
Daisy 1.0.8.32 build 149 — GPT-6.1 Sol default, opt-in crash reports, pause alignment
GPT-6.1 Sol becomes the default OpenAI summary model, replacing GPT-5.6 Terra which stays selectable. Crash reports are opt-in and shown in full first, and recordings with pauses keep both voices aligned in exports.
Daisy 1.0.8.31 build 148 — mic and system audio start on one clock
Microphone and system-audio recordings now share a start time, so mixed and two-track exports no longer drift. Recordings made before this version keep their original timing.
Daisy 1.0.8.30 build 147 — two-track audio export
Export audio can write your voice and the other side to separate files with a Two tracks option, alongside the existing One file mode. Also fixes garbled Cyrillic when pasting into TextEdit, Word or Pages.
Daisy 1.0.8.27 build 144 — processing plans in onboarding
Onboarding adds a step to pick Economy, Balanced or Maximum processing and whether to defer processing until the Mac is on a charger. Both can be changed later in Settings → General.
Daisy 1.0.8.24 build 141 — streaming cloud summaries and MCP without Node.js
Cloud summaries from Anthropic, OpenAI, Kimi and Gemini now stream so long meetings no longer time out. MCP connections no longer need Node.js, and Gemini joins as a summary provider.
What people actually say about Daisy — is it worth it?
We scanned public community sources for Daisy on Sep 1, 2026 and could not establish that the discussion we found is about this tool rather than something else sharing its name. Our own analysis of that scan says the posts were off-subject. Rather than publish a sentiment score built on the wrong subject, we publish nothing here and re-run the scan.
Viability Score
How well maintained and how widely used is Daisy? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- Meetings mode records microphone plus Mac system audio with no bot in the call
- On-device transcription in 12 languages with auto-detect
- Speaker separation with 'Same person as…' merge and undo
- Push-to-talk dictation pastes into any app and saves nothing
- Tap-to-start voice notes saved as titled, searchable standalone notes
- Markdown transcript archive in any folder, Obsidian-friendly
- Project-aware meeting summaries that read earlier notes and summaries for context
- Meetings auto-title themselves
- Next steps filed to Notion, Linear, Jira or GitHub, or emailed to attendees
- MCP server exposes the local archive to Claude Desktop, Cursor, Cline and Continue over a loopback bridge with no Node.js
- Meeting preparation with agendas imported from text, Markdown, PDF or DOCX
- Evidence-backed plan review with transcript timestamps per agenda item
- Two-track audio export writing your voice and the other side to separate files that share one clock
- Processing plans: Economy, Balanced and Maximum, with deferred processing while on a charger
- Optional local privacy layer pseudonymizes people and companies and redacts card numbers
About Daisy
Daisy is a native macOS app with three capture modes in one install. Meetings records your microphone plus whatever plays on your Mac — no bot joins the call — separates speakers and turns the call into a finished note. Dictation is hold-a-hotkey push-to-talk that pastes into whatever app has your cursor and saves nothing. Voice notes turn a spoken thought into a titled, searchable note in your library. Transcription runs on-device in 12 languages with auto-detect. The current 1.0.8 line (1.0.8.32, build 149, 10 Oct 2026) is where the workflow tightened. Project-aware summaries read a project's earlier notes for context, and meetings title themselves. Next steps can be filed to Notion, Linear, Jira or GitHub, or emailed to attendees. Meeting prep imports agendas from text, Markdown, PDF or DOCX, and evidence-backed plan review walks each agenda item against transcript timestamps. Onboarding offers Economy, Balanced and Maximum processing plans, plus the option to defer processing until your Mac is on a charger. Export audio can write your voice and the other side to two separate tracks that now start on one clock. Ownership is the design principle. Transcripts are plain Markdown in a folder you choose — Obsidian vaults included — and an MCP server exposes that archive to Claude Desktop, Cursor, Cline and Continue over a loopback connection, with Claude Code connecting in one click and no Node.js required. Summaries can run through Anthropic (Claude Sonnet 5.5 recommended), OpenAI (GPT-6.1 Sol is the current default), Kimi, Gemini, or Gemma 4 via Ollama and LM Studio, with an optional local privacy layer that pseudonymizes people and companies while card numbers stay redacted. Daisy is Apache 2.0 open source and needs no Daisy account. Against Otter, Fireflies and Granola, the trade is custody versus collaboration. Those bring cloud workspaces, shared transcripts and admin controls; Daisy is Mac-only with no team spaces and no web app.
Behind the Verdict
Daisy's bet is that the recording of a conversation is a document, not a service. Three modes share one widget: Meetings captures mic plus system audio with no bot in the call, Dictation is a hold-to-talk paste that saves nothing, Voice notes drop a titled note into your library. All three land in a Markdown folder you pick, which is why Obsidian users get transcripts beside their existing notes. The workflow layer is where it earns its keep beyond transcription. Project-aware summaries read a project's earlier notes and summaries for context; meeting prep imports agendas from text, Markdown, PDF or DOCX; evidence-backed plan review walks each agenda item against transcript timestamps, so a plan gets checked against what was actually said rather than memory. Next steps push out to Notion, Linear, Jira or GitHub, or go straight to attendees by email. The MCP server is the sleeper feature — Claude Desktop, Cursor, Cline and Continue can query your session archive over a loopback connection, and 1.0.8.24 removed the Node.js dependency so Claude and Codex connect through a built-in bridge and Claude Code connects in one click. Summaries are the one place a cloud boundary exists, and Daisy is unusually honest about it. In 1.0.8.26 the summary providers are Claude Sonnet 5.5 (recommended), Opus 5.5, Fable 5.1 and Haiku 5.5; OpenAI GPT-6.1 Sol, GPT-6 Astra and GPT-6 Luna; Kimi; Gemini; and Gemma 4 for Ollama and LM Studio. GPT-6.1 Sol became the default in 1.0.8.32 as faster and cheaper than GPT-5.6 Terra, which stays selectable. You can also route summaries through a ChatGPT account already signed in through ChatGPT or Codex, subject to that plan's limits. A local privacy layer pseudonymizes people and companies and redacts card numbers before anything leaves. Cloud summaries stream as of 1.0.8.24, so a two-hour meeting no longer times out and isn't billed twice on retry. Where it doesn't fit: Windows and Linux users have no build; teams that need shared workspaces, role-based access or collaborative transcript editing will not find them; anyone who wants a web interface or cross-device sync of the archive is in the wrong product. The release cadence — multiple builds dated the same day in early October 2026 — is either a virtue or a tax depending on your tolerance. And because it's a signed DMG requiring system permissions on first launch, first-run friction is real, though the docs walk installation, first recording and modes in short pages.
Researching Daisy? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Daisy actually fits — and what changes day-one when you adopt it.
Start a client call with Meetings mode, which grabs mic plus system audio with no bot; Daisy separates speakers and writes a Markdown note into an Obsidian vault folder.
Outcome: A finished, titled transcript on your own disk before the call even ends, with next steps filed to Linear or emailed to the client.
Point the MCP server at the local session archive — Claude Code connects in one click, with no Node.js install — then ask the client to pull decisions from last week's sessions.
Outcome: Meeting history becomes queryable context for your coding assistant without uploading transcripts anywhere.
Import a DOCX agenda into meeting prep, record the interview, then run evidence-backed plan review that checks each agenda item against transcript timestamps.
Outcome: A per-item audit trail with timestamps, and summaries pseudonymizing names and companies before any cloud call.
Use Cases
- Record and transcribe remote meetings without a bot joining or a cloud upload
- Use push-to-talk dictation to type anywhere on your Mac with local transcription
- Expose meeting transcripts to Claude Desktop, Cursor or Claude Code via the MCP server
- Transcribe existing audio and video files offline
- Keep transcripts as Markdown in an Obsidian vault or iCloud Drive folder
- File meeting next steps straight to Notion, Linear, Jira or GitHub
- Import a PDF or DOCX agenda and review the plan against transcript timestamps
- Export your voice and the other side as two aligned audio tracks
Models Under the Hood
as of 2026-10-15
Limitations
- Daisy is a signed macOS DMG and asks for system permissions on first launch, so first-run setup is not zero-touch.
- Transcription is on-device, but summaries can cross a cloud boundary: the docs describe using a ChatGPT account already signed in through ChatGPT or Codex with that plan's limits applying, or a key for Anthropic, OpenAI, Kimi or Gemini.
- A missing summary has documented causes — Apple Intelligence switched off, a language it doesn't write yet, a provider error — and the meeting keeps a note with Try again.
- MCP sessions are exposed over a loopback connection only.
- The changelog shows a fast cadence, with several 1.0.8 builds dated 9 October 2026.
as of 2026-10-17
Verification history
We have re-verified Daisy 9 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 9 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Daisy tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Beta
$0
Ideal for
Mac users who want on-device meeting capture, dictation and voice notes with a Markdown archive, paying only their own summary provider's token costs
What this tier adds
Starting tier: meetings capture with no bot, on-device transcription in 12 languages, push-to-talk dictation, voice notes, Markdown archive, Apache 2.0 source, no Daisy account
Where the pricing makes sense
The company stage and team size where Daisy's pricing actually pencils out — and where peers do it cheaper.
Daisy runs on the ChatGPT, Anthropic, OpenAI, Kimi or Gemini account you already pay for, or on local Gemma 4 through Ollama and LM Studio at no model cost — so the marginal cost of a meeting is tokens, not a per-seat subscription. Compared with Otter or Fireflies, which bundle cloud storage and seats into a team plan, Daisy's spend scales with how much you summarize rather than how many people you add.
Setup time & first value
How long it actually takes to get something useful out of Daisy — broken out by persona, not the marketing-page minute.
Solo Mac user: 10–15 minutes from DMG to a first summarized meeting, counting permission grants and the 1.0.8.27 processing-plan step. Developer wiring MCP into Claude Code or Cursor: add 5 minutes, since 1.0.8.24 removed the Node.js requirement. Anyone connecting a cloud summary key should budget another 5.
Switching to or from Daisy
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Apple Voice Memos: import the audio files and transcribe them offline inside Daisy
- →From Otter: export your existing transcripts, then start new recordings in Daisy and keep the archive as Markdown in a folder you choose
- →From Granola: keep Granola's cloud notes where they are and route new Mac recordings through Daisy into an Obsidian vault
- →From Wispr Flow: move push-to-talk dictation to Daisy's hold-a-hotkey mode, which pastes into any app and saves nothing
- ↗To Otter or Fireflies: take the Markdown transcripts with you; you lose on-device capture and the MCP bridge but gain shared workspaces
- ↗To Granola: useful if you want cloud notes plus a Mac app, at the cost of transcripts living on someone else's server
- ↗To Apple's built-in transcription: cheaper but drops speaker separation, project-aware summaries and the MCP archive
Integrations
Resources & Guides
- Documentationmydaisy.io
Docs · Daisy
Full product docs from mydaisy.io
- Quickstartmydaisy.io
Installation · Daisy
Get up and running fast from mydaisy.io
- Quickstartmydaisy.io
First Recording · Daisy
Get up and running fast from mydaisy.io
- Documentationmydaisy.io
Chatgpt Account · Daisy
Full product docs from mydaisy.io
- Resourcemydaisy.io
Changelog · Daisy
Helpful link from mydaisy.io
Tutorials & Learning
YouTube returned 6 videos for “Daisy”, and we withheld 6: 6 could not be judged, because “Daisy” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Daisy.
Official links
Tools that pair well with Daisy
Common stack mates teams adopt alongside Daisy, with the specific reason each pairing earns its keep.
Voicenotes
Voicenotes is a bot-free AI notetaker that records meetings on your own device and hands you the summary before you hang up.
Jamie
Bot-free AI meeting notes that record from your own device, not a bot in the call.
Plaud
Plaud pairs wearable AI note takers with transcription, AI Notes, and a voice-driven Plaud Agent.
Featured Head-to-Head Comparisons
Daisy vs Poke Interaction Co
Choose Daisy if you're a Mac user who values absolute privacy and needs local meeting transcription, dictation, and markdown workflow integration — it's free and open-source. Pick Poke if you want an AI assistant that handles email, calendar, tasks, and health data directly from your messaging apps, and you're okay with a cloud subscription model.
Daisy vs Gem
Gem and Daisy serve completely different markets: Gem is a high-cost recruiting platform for enterprises, while Daisy is a free, privacy-first local transcription tool for Mac users. Choose Gem if you run a recruiting team and need an all-in-one ATS+CRM with AI agents; choose Daisy if you want secure, offline meeting recording and dictation without cloud dependency.
Daisy vs Guesty
Guesty and Daisy serve completely different needs. Guesty is a full-fledged PMS for vacation rental operators, with AI that automates guest messaging and reconciliation. Daisy is a local-first meeting recorder for Mac users who want privacy and on-device AI. Your choice depends entirely on whether you manage properties or need to transcribe meetings without cloud dependency.
Alternatives to Daisy
View all
Voicenotes
Voicenotes is a bot-free AI notetaker that records meetings on your own device and hands you the summary before you hang up.
Frequently Asked Questions
Best-of guides
Used Daisy? Help shape our editorial sentiment research.