Frequently Asked Questions

Yes. Recording, transcription, speaker labels, and search all run on your machine, and Daisy is fully functional offline. Only two things ever reach the network on their own — both to our own daisy.smbr.app, both carrying no meeting content, and both fail-silent:

• An update check — notify-only, on by default but easy to turn off (setup wizard, or Settings → Recordings). It tells you a new version exists and never installs anything on its own.
• A license check-in — at most once a day, to confirm your license is valid.

Everything else happens only because you set it up: an AI provider with your own key (or a fully local model), a calendar you subscribe to, a webhook you add, or a model you download. Help and changelog links open in your browser when you click them. The app window itself cannot make network calls at all — a strict content-security policy blocks it.

Full Security & Threat Model →

Yes, with the right configuration. Because recording, transcription, and speaker labels run on your machine, Daisy can be set up so protected health information never leaves your device — which keeps you the only party holding it. Small Bricktory is not your business associate and offers no BAA: in the local setup nothing PHI-bearing reaches us. If you use a cloud AI model, your BAA is directly with that provider.

Daisy is not “HIPAA certified” — no such certification exists for software. See Configuring Daisy for HIPAA and PHI for the setup mapped to each Security Rule safeguard.

Yes — completely. Daisy never dials a participant into your call. It records the meeting the way your own ears hear it: your microphone and your computer's audio output, captured as two separate channels on your machine. Nobody sees a notetaker in the participant list, because there isn't one to see.

That matters more every quarter. Microsoft Teams now detects third-party recording bots, parks them in a Suspected threats lobby marked “Unverified,” and makes the organizer admit them with a deliberate extra click — and as of August 2026, admins are getting a switch to auto-block every external AI bot outright. Google Meet and Zoom are tightening the same way. If your notetaker works by joining the call, it can be flagged, held in a lobby, or shut off by an IT policy you don't control — usually on someone else's tenant, in the meeting that mattered. Daisy has no bot to flag, no bot to admit, and no bot to block.

The bot is only half of it. What lasts is what happens after capture: cloud notetakers still upload your audio to understand it. Daisy transcribes, labels speakers, and searches on your own hardware, so the recording never leaves the room it was made in.

See the side-by-side comparison →

Yes — a full, code-sourced writeup: the local-first architecture, the encrypted credential vault, a complete inventory of what leaves your machine and when, the threats it defends (and the ones it doesn't), and what you should do.

Security & Threat Model →

Onto your own disk, and nowhere else by default. Recordings live as files in a folder you choose (.wav chunks plus a compact stereo .opus archive). Recording, transcription, speaker identification, and search all run on your machine.

The only ways anything leaves are ones you turn on. You press “Send to…”. You opt into a cloud model for the AI step with your own key — there only the transcript text goes, over HTTPS, never the audio. Transcription, live captions, and speaker labels stay entirely on your machine, always. Private by default, your choice to extend.

Your recordings, transcripts, summaries, and notes stay on your disk in a folder you choose. Nothing is uploaded unless you explicitly send it or pick a cloud provider.

Of course, if you keep that folder in a cloud-synced location — Dropbox, OneDrive, Syncthing — every device can share the same profile and recordings, but those files now also live on that cloud. Your call, and out of our hands.

Mostly. Recording, transcription, speaker labels, and search work with the network off, and they keep working through extended periods offline. Summaries and Q&A need a language model — run a local one (Ollama or LM Studio) and even those stay fully local; bring a cloud key and only the transcript text goes out when you have a connection.

In the spirit of transparency — we don't want your data — here is every time Daisy touches the network:

• It checks your license key's validity at most once a day.
• Update checks and device activation.

That's the whole list. Your meetings, audio, and transcripts are never part of any of it.

Daisy writes a quiet log file to your profile folder (logs/daisy.log) with stage timings and error context — enough to pin down a slow finalize or an unhappy mic without ever leaving your machine. Logs rotate daily; the last 7 days are kept and older rotations are deleted.

Nothing ships anywhere automatically. There's an Open folder button under Settings → Recordings → Log files, plus a Detailed logs setting if you're chasing something specific. Diagnostics & logs.

The everyday proof is the design itself: Daisy is local-first, so your recordings, transcripts, and summaries sit on your own disk in open file formats (.wav, .opus, plain text) that you can read and move without us. You don't have to take our word for where your data is — you can see it.

Beyond that, you don't have to trust us at all: Daisy is source-available. The whole client — recording, transcription, the credential vault, and every network call it makes — is published at github.com/smallbricktory/daisy-local. Read it, audit it, or build it yourself — no NDA, no fee, no request form. If anything on this page isn't backed by the code, that's a bug we want to hear about. See the Security & Threat Model for the full inventory of what leaves your machine.

When you reclaim space under Settings → Recordings, Daisy removes the bulky raw WAVs but keeps a compact stereo meeting.opus archive (your mic on one channel, the system audio on the other). You can re-transcribe an old meeting later — or re-label the speakers — from that archive, best-effort.

Deleting a meeting outright from the Library wipes the Opus archive too — that one is final.

Those tools process your audio in their cloud, store your transcripts under their terms, and run them through a model you don't control. Several now offer a bot-free capture mode, which changes who sees a notetaker in the participant list — it doesn't change where the audio goes.

Daisy keeps recording, transcription, and speaker identification on your machine, with no bot and no account, and lets you bring your own model or key. The difference isn't the bot; it's that nothing is uploaded to be understood in the first place.

See the side-by-side comparison →

Most meeting tools work the other way around: a bot joins your call, your audio is streamed to a company's servers, processed there, and your transcripts live in an account you don't control. Daisy flips that — it's an app on your computer, and your recordings, transcripts, summaries, and notes are saved as files on your disk in a folder you choose.

Recording, transcription, speaker labels, and search all run on-device and keep working offline. Anything online — a cloud AI key for summaries, a calendar you subscribe to, a webhook — is something you opt into, never the default. Your stuff lives with you first.

Mostly yes. Recording, transcription, and speaker identification all run on your machine out of the box — the speech-to-text model is bundled, no download or key needed.

Summaries are the one piece that always needs an AI service. You can run it two ways: your own API key with a cloud provider (Anthropic, OpenAI, or Groq), or — if you want fully offline — a local AI server like LM Studio or Ollama, the advanced path. Or skip the provider entirely and copy the transcript into your own ChatGPT for free.

Daisy identifies speakers right on your device. Label someone once and it will recognize their voice in future meetings. Voices are encrypted, local to your machine, and never uploaded to us or anyone.

In long meetings where a person's voice drifts (they move around, switch rooms, get tired), Daisy collapses what would otherwise be two clusters back into one identity as long as both match the same vault voiceprint. You don't have to merge by hand.

Your finished transcript and speaker labels are free and produced on your machine — no per-minute bill or API key.

The summary and other AI features are where a model comes in. Run a local model (LM Studio or Ollama) and it's free, on your own hardware. Or use a cloud key: a full meeting's AI — summary, chapters, and analysis — works out to roughly 4¢ on a balanced mix of efficient models, up to about 30¢ if you force a top-tier model for everything. Those are approximate and vary by provider; the charge goes straight to your own key — Daisy never proxies or marks up. You can also skip it and copy-paste the transcript into your own ChatGPT or Claude, free.

On-device, always, on every machine — and there is no configuration in which that changes. The cloud never touches transcription: live captions, the finished transcript, and speaker labels are all produced locally and free.

They're also not a separate feature. Daisy transcribes while you record and keeps that text when you stop, so the captions you watch appear are the transcript. The switch under Settings → Recordings → Live captions only decides whether the window shows them to you as they arrive; turning it off doesn't change the transcript, and doesn't skip any work.

One of three, all running on your own machine, listed together under Settings → Providers → Local transcription:

• Windows and Linux — NVIDIA's Parakeet (TDT 0.6B v2), bundled inside the app. English, CPU-only, no GPU needed, and nothing to download: the first recording works straight out of the installer.
• macOS 26 and later — Apple's own on-device speech engine, using the speech assets your Mac already manages. Covers the languages your system settings do.
• Whisper — ten models from Tiny to Large v3, downloadable on any platform. This is the multilingual path, the choice for older Macs, and the fallback Daisy drops to if the engine you picked won't start.

Every model Daisy downloads is SHA-256-verified against a pin inside the binary before it's used. Switching models takes effect on your next recording.

The recording keeps going. Daisy watches both audio streams for the length of a meeting: if your microphone disappears it moves to another physical one and tells you which, and if system audio drops out it reconnects. Virtual devices are never picked for you.

The cases it can't fix, it names instead of hiding — a mic muted in the system mixer, an input another app has taken over, a device streaming silence. Each is a corner toast saying what happened and whether Daisy is recovering on its own, and each is written to the log. Microphones and audio devices.

Recordings rotate into bounded chunks on disk every few minutes, so an unexpected crash mid-meeting costs you at most the open chunk — not the whole session. Reopen Daisy and the in-progress session is recovered automatically, and a chunk that was mid-write when the power went is salvaged up to its last valid audio rather than discarded.

If the app window goes away while a meeting is finalizing, a background process keeps the local work going. When you reopen Daisy, a corner toast catches you up: “Finalized” with a click-through to the session.

It's expected. Daisy records two independent streams — your microphone (Me) and your system audio (Them) — and your mic can pick up the other person's voice through your speakers, so the same sentence gets transcribed on both sides.

You don't need to do anything. When the call ends, Daisy reprocesses the recording, matches it against your voiceprints to label each speaker, and merges those duplicates into a single attributed turn. The saved transcript is the source of truth; the live view is a preview. Wearing headphones avoids it almost entirely. More on live transcription.

No. Recording, transcription, and speaker identification all run on your machine. You can use Daisy day-to-day with no key and no account.

You only need a key if you want Daisy to write summaries, answer questions across your meetings, generate chapters, or run an analysis for you. Even then, you can skip the key entirely and use the Copy all button on a transcript to paste it into your own ChatGPT or Claude.

First, you have two ways to run the AI step at all: a local model on your own hardware (nothing leaves), or your own cloud key. This question is about the cloud-key route.

Daisy supports Anthropic (Claude), OpenAI (GPT), and Groq. Any of the three works for summaries — pick whichever you already have an account with. Groq has a usable free tier if you want to start without a card.

You only need one. You can switch later without losing anything. Step-by-step setup for each provider.

Open Daisy and go to Settings → Providers. Pick the provider you generated a key for under AI provider, click Configure, paste the key, click Test key & list models →, pick a model, and Save.

Daisy stores the key in your local encrypted vault — the same vault that holds your voiceprints — and never sends it anywhere except the provider you picked.

Pennies. A full meeting's AI — summary, chapters, and analysis — works out to roughly 4¢ for a 1-hour meeting on a balanced mix of efficient models, and up to about 30¢ if you force a top-tier model for everything. Those are approximate and vary by provider; the charge goes straight to your own key, and Daisy never proxies or marks up.

As soon as your key works, hop back to the provider's console and set a monthly spending cap (Anthropic: Settings → Limits; OpenAI: Settings → Limits; Groq: Settings → Billing) so a runaway bill is impossible. A $10–$25 monthly cap is plenty for solo use. The price of Granola or Otter — but with all of the control.

For practical purposes, it's already done when the meeting ends. Daisy transcribes as people speak and keeps that text rather than transcribing the audio a second time, so the transcription cost is spread across the call instead of landing in a lump at the end.

The default engine on each platform is picked to make that affordable: a CPU-only model bundled with the Windows and Linux builds, and Apple's own on-device engine on a recent Mac. Neither needs a GPU. What still runs after you press Finish is echo cancellation, speaker labels, and the summary — plus a catch-up pass over any stretch the live pass didn't cover, which on a normal meeting is nothing.

A full transcription pass only happens in two cases: audio you imported (there was no live pass), and a session recorded with no model installed or on a machine that couldn't keep up. You'll see a “transcribed n/m chunks” progress bar when that's happening. That's the case where a GPU and a smaller model both help.

The AI features (summary, chapters, analysis, asking questions) are the part that takes a moment. On a capable GPU a local model writes an hour-long meeting's summary in roughly 15–20 seconds; on a CPU-only machine the same work can stretch to a few minutes. A cloud key returns it in seconds regardless. It all runs in the background, so you can close the app while it finishes.

Yes — end a search with a question mark and Daisy answers from across your meetings, with citations linking back to the exact moment each point came from.

Unlike keyword search, asking questions needs a language model, because Daisy has to read and reason over your transcripts to compose the answer. That can be a local model (so nothing leaves your machine) or a cloud key you bring — the same provider setting your summaries use.

Keyword search — yes, completely. It's fully local, with no AI and no network.

Asking questions needs a language model. With a cloud key that step needs internet; with a local model it works fully offline. Install LM Studio or Ollama and run a compact model — something like Gemma 3n E4B is small enough to run on a MacBook — and you can ask questions across your whole history with the network off.

Workflows are Daisy's built-in automation. A workflow is a rule: when something happens to a recording (it's finalized, deleted, imported, or fails), if it matches your conditions (tag, participant, or title), do one or more actions — run an Analyzer prompt, or send the meeting to an integration. Rules run locally in the background; building one sends nothing off your machine.

Workflows: automate what happens to a recording →

Yes. Add a webhook destination in Settings → Integrations (an https:// URL, with optional header or bearer auth), choose whether to include the summary, notes, and/or transcript, then use ↗ Send To… in a meeting's Action menu — or let a Workflow do it automatically. Slack, Discord, n8n, Zapier, Make, and your own endpoint all just receive a single JSON POST; there's no connector to install.

Integrations: send your meetings anywhere →

Yes — 30 days, free, with no account, no card, and no email. Download Daisy for your platform, install it, and the trial starts the first time you open it; there's no sign-up step. Record, transcribe, and summarize as much as you like for the full 30 days, and your recordings and transcripts stay on your own machine the whole time.

When you're ready, pick a plan on the pricing page and paste the emailed key under Settings → About — it activates on up to 3 of your own devices.

You buy a license one of two ways: a subscription (monthly or yearly) or a one-time perpetual license. Either way there is no account — we email you a signed license key and you paste it in once under Settings → About. That screen shows who the app is licensed to and a short License ID fingerprint rather than the full key.

One license covers up to 3 of your own devices. Daisy verifies the key on your device and does a light validity check-in at most once a day; in between — and through long stretches offline — it keeps working. Move a seat any time with Deactivate this device, which frees it for another machine. Your recordings live in a folder on your disk and stay yours no matter what your license does.

On a subscription, the clock starts when the key is generated at purchase, not when you first activate it — so activate soon after buying. Daisy quietly refreshes the key in the background before it expires, so you never re-paste; each renewal also emails a fresh key as a backup. A perpetual license never expires and has nothing to renew.

Yes — Daisy is source-available. Read, audit, modify, and build it yourself at github.com/smallbricktory/daisy-local, under the Elastic License 2.0.

In the spirit of building to share, self-builds are permitted, and bypassing the licensing code is permitted for personal, non-commercial use only. Commercial use requires a paid license — which also funds development and the signed, notarized builds most people run.

Yes. The full client lives at github.com/smallbricktory/daisy-local. Clone it and follow the build steps in the README — it's a Rust workspace with a Tauri front-end, so you'll need the Rust toolchain and Node installed. What you build is the same app we ship.

• Personal, non-commercial use: you may build it and bypass the licensing code — no key required.
• Commercial use: needs a paid license, which also funds development and the signed builds most people run.
• A self-build isn't code-signed or notarized by us, so your operating system may warn you before it runs the first time. Updates are yours to pull and rebuild.

Since Daisy never took your email, there's no reminder in your inbox — so this is the path back. Your meetings are still on your disk in the folder you chose; they were never on our servers. Re-download Daisy, point it at that folder, and everything is where you left it.

Pick up where you left off →