Best AI Note Taker for In-Person Meetings (2026)

The best AI note taker for an in-person meeting is one that records without a bot — because no bot can join a conversation around a table. In 2026 that means either a dedicated recording device (Plaud, Kuno) or a phone/desktop app that captures your microphone directly (Fellow, Otter, Jamie, Krisp, Fireflies). Pick based on room size, privacy, and where your audio is processed.

🎙️ Quick answer: For team workflows, Fellow captures in-person audio bot-free and routes action items into your tools. For dedicated hardware, Plaud leads on form factor. For EU privacy and data sovereignty, look at Jamie (EU-hosted software) and Kuno (a recording device made in Germany with on-device transcription). The right pick depends on three things: room size, what happens to your audio, and your team’s privacy rules.

In-person meetings are still where the important decisions happen — roughly 44% of meetings in Europe are held face-to-face, and the formats most often run in person are all-hands, one-on-ones, and onboarding (according to 2025 workplace-meeting statistics roundups). Yet almost every popular AI note taker was built to join a Zoom, Teams, or Meet call as a bot — a workflow that simply breaks the moment you put your laptop away and sit down across a table. (Stat source: Archie 2025 meeting statistics. Verified June 2026.)

✅ How we picked. Each tool below is judged on four things that actually matter in a physical room: (1) bot-free capture, (2) mic range / audio quality across a table, (3) offline reliability (client Wi‑Fi is often locked down), and (4) where your audio is processed and stored — the privacy dimension most lists skip.

Quick comparison: the 7 best in-person AI note takers

Tool	Type	In-person capture	Best for	Key limitation
Fellow	Software (desktop + mobile)	Bot-free app recording	Teams that need action items in Slack/Jira/HubSpot	Workspace tool — heavy for a solo user
Plaud (Note / Note Pro / NotePin)	Hardware device	On-device recorder, syncs later	Boardrooms, polished client settings	Cloud processing + paid plan for real usage
Otter.ai	Software (mobile)	Live phone-mic transcript	Real-time transcript you can watch	Phone mic struggles in big/noisy rooms
Jamie	Software (desktop)	Bot-free, EU-hosted	European teams that need GDPR + EU data residency	Narrower integrations ecosystem
Krisp	Software (system audio)	Records any device audio	Loud rooms / open offices (noise cancellation)	Note-taking features newer than its core
Fireflies	Software (mobile + bot)	Mobile app for in-person	Multilingual teams (100+ languages)	Built primarily for video calls
Kuno	Hardware device (made in Germany)	On-device transcription, EU-hosted	Privacy-first capture of field/in-person meetings	Pre-launch (early access)

The 7 best AI note takers for in-person meetings

1. Fellow — best overall for teams

Fellow records in-person meetings bot-free through its desktop and mobile apps, then generates structured summaries and assigns action items to owners. Its strength is what happens after the meeting: outputs sync to Slack, HubSpot, Jira and 50+ tools, and speaker diarization labels who said what even with several voices in the room. It’s a workspace tool, so a solo user with no workflow needs may find it heavier than necessary. Pricing: free plan; Pro from about $7/user/month. (Source: fellow.ai.)

2. Plaud — best dedicated hardware

Plaud is the hardware-first option: a credit-card-sized recorder (Note, Note Pro) or a wearable pin (NotePin, NotePin S) that records locally and processes with AI in the Plaud app after you sync. The Note Pro stores audio on 64 GB onboard, runs offline during the meeting, and shows recording status on a small screen — useful when you set it on a client’s table. The trade-off: transcription and summaries are produced in Plaud’s cloud after upload, and meaningful use needs a paid plan on top of the device. Pricing (verified June 2026): Plaud Note $159, Note Pro $189, NotePin $159, NotePin S $179; free Starter plan (300 min/month), Pro $99.99/year (1,200 min/month), Unlimited $239.99/year. (Source: plaud.ai.)

3. Otter.ai — best for a live transcript

Otter streams a transcript to your phone screen as people speak, which is genuinely handy in person when you want to catch a misheard name or figure without replaying audio. Summaries and action items come after the meeting, and Otter Chat lets you query the transcript. The weak link in a physical room is the phone microphone: someone across a large table can get garbled, and real-time features lean on a connection. Pricing (verified June 2026): free plan; Pro from $8.33/user/month billed annually; Business from $19.99/user/month. (Source: otter.ai/pricing.)

4. Jamie — best EU-hosted software option

Jamie is built around the premise that your meeting data shouldn’t leave the EU. It’s bot-free, runs from a desktop app for in-person, hybrid and virtual meetings, and — per its own site — stores everything encrypted on EU servers, is GDPR compliant and ISO 27001 certified, and deletes the source audio once the summary is ready. That makes it a strong software choice for European teams. Integrations are narrower than Fellow’s, so CRM/PM hand-off may need Zapier. Pricing: free plan; paid tiers priced in EUR (around €21+/month as reported — verify current pricing). (Source: meetjamie.ai.)

5. Krisp — best for noisy rooms

Krisp began as noise cancellation and still leads there: it cleans the audio signal before transcription, so open offices, café tables and HVAC hum hurt the transcript less. It works at the system-audio level across any app. The note-taking layer is newer and less mature than its core, and deep workflow/governance controls are limited. Pricing: free plan (with noise cancellation); Pro around $16/month as reported — verify current pricing. (Source: fellow.ai.)

6. Fireflies — best for multilingual teams

Fireflies is primarily a video-call notetaker, but its mobile app records in-person audio and it supports 100+ languages — useful for multilingual or cross-border teams. Security-conscious organizations have flagged its unauthenticated sharing links and prompts that ask external participants to sign up, so verify its posture before deploying in a regulated setting. Pricing: free plan; Pro around $18/user/month, Business around $29/user/month as reported — verify current pricing. (Source: fellow.ai.)

7. Kuno — best for privacy-first, field & in-person capture (made in Germany)

Kuno is a dedicated recording device made in Germany aimed squarely at the in-person and field meetings software bots can’t reach. What sets it apart on this list is on-device transcription — the audio is processed on the device, so it doesn’t have to be shipped to a US cloud — combined with EU hosting and a commitment that recordings are never used to train AI. For the consent problem that in-person recording creates, Kuno shows a visible recording indicator so everyone in the room sees when it’s on, and a one-tap physical stop switch to pause instantly. It’s pre-launch, so access is via early access rather than purchase today.

▶️ Capture in-person meetings without shipping them to a US cloud. Kuno records and transcribes your face-to-face and field meetings on a device made in Germany — on-device transcription so audio never leaves the room, EU-hosted, and never used to train AI. A visible recording indicator and one-tap stop keep consent clean. Get early access → (Pre-launch; product page follows at launch.)

How much do AI note takers for in-person meetings cost?

Hardware is a one-time cost plus a plan; software is a subscription. Here’s the 2026 picture, with vendor-verified figures where available.

Tool	Up-front	Subscription	Free tier?
Fellow	—	Pro from ~$7/user/mo	Yes
Plaud Note / Note Pro	$159 / $189 device	Free 300 min/mo · Pro $99.99/yr · Unlimited $239.99/yr	Yes (Starter)
Otter.ai	—	Pro from $8.33/user/mo (annual) · Business from $19.99/user/mo	Yes
Jamie	—	Paid from ~€21/mo (EUR) — verify	Yes
Krisp	—	Pro ~$16/mo — verify	Yes
Fireflies	—	Pro ~$18/user/mo · Business ~$29/user/mo — verify	Yes
Kuno	Device (early access)	TBA at launch	Early access

⚠️ Verify before you buy. Device prices (Plaud) and Otter’s tiers are taken directly from the vendors’ own pages (June 2026). Figures marked “verify” are as reported by third-party comparisons and change often — confirm on the vendor’s pricing page before purchasing.

Where does your meeting audio actually go? (the question lists skip)

This is the dimension most “best in-person note taker” articles flatten — and it’s the one that decides whether a tool is allowed in a regulated team at all. “Records on-device” is not the same as “processes on-device.” Many hardware recorders capture locally but still upload your audio to a cloud to transcribe and summarize it. Where that cloud sits — and whether your recording is used to train a model — is the real privacy question.

Tool	Capture	Transcription / AI processing	Data-sovereignty note
Fellow	Device mic (app)	Cloud (enterprise security controls)	Team governance focus; check region
Plaud	On-device recording	Cloud (after sync)	Audio leaves the device for processing
Otter.ai	Phone mic (app)	Cloud (US)	US-hosted
Jamie	Device mic (app)	Cloud, EU servers; audio deleted after summary	EU-hosted, GDPR, ISO 27001
Krisp	System audio	On-device noise cancel; notes in cloud	Check region for note processing
Fireflies	Phone mic (app)	Cloud	Verify sharing/security posture
Kuno	On-device recording	On-device transcription, EU-hosted	Made in Germany; no training on your recordings

📌 Why this matters in 2026. When Limitless was acquired by Meta in December 2025, it stopped selling its Pendant wearable and pulled service in several markets — EU, UK and others — giving affected users until 19 December 2025 to export data (Neowin; 9to5Mac). For anyone whose notes contain client or patient detail, where the data lives — and who owns the company holding it — is not a footnote. (Verified June 2026.)

Can an AI note taker work without a bot?

Yes — and for in-person meetings it has to. Bot-free tools capture audio directly through your device’s microphone (or a dedicated recorder) instead of joining a call as a participant. Fellow, Jamie, Krisp and Granola are all bot-free; Plaud and Kuno skip software entirely with a physical device. There’s no meeting link, no “recording bot has joined,” and nothing for a physical room to reject.

How does an AI note taker record an in-person meeting?

Three mechanisms dominate. A mobile or desktop app uses your device microphone, then transcribes and summarizes. A dedicated hardware recorder captures locally and processes afterward (some, like Kuno, transcribe on the device itself). A phone voice memo + upload workflow is the zero-cost version: record, then drop the file into a transcription tool. The differences that bite are mic range across a table and whether processing needs the internet.

Hardware vs software vs phone: which is best for in-person?

It depends on room size and stakes. Software apps are free to try and always with you, but a phone mic fades past three to four feet and can look unprofessional on a client table. Dedicated hardware picks up a whole room, runs offline, and looks intentional — at the cost of a device plus a plan. Your phone’s recorder is fine for informal internal chats and terrible for a boardroom.

Approach	Room coverage	Offline?	Looks professional?	Best when
Software app (phone/desktop)	Small rooms, near the mic	Limited (live features need a connection)	Mixed (phone on table)	You already pay for the tool; small meetings
Dedicated hardware	Whole table / boardroom	Yes (records locally)	Yes	Client-facing, larger or sensitive meetings
Phone voice memo + upload	Very short range	Yes (record) / no (processing)	No	Informal internal notes, zero budget

Do AI note takers work offline?

Recording can; processing usually can’t. Hardware recorders (Plaud, Kuno) store audio locally with no Wi‑Fi, which is exactly why they shine in client offices with locked-down guest networks. The AI summary then needs compute — in the cloud for most tools, or on the device itself for on-device transcription. Pure software apps that stream a live transcript depend on a connection for the real-time part, though many can still record offline and process later.

Which AI note taker is best for noisy rooms?

Krisp, because its noise cancellation runs before transcription and cleans the signal at the source — a real edge in open offices and cafés. Dedicated hardware with a multi-microphone array (for example Plaud’s four-mic Note Pro) also handles larger, busier rooms better than a single phone mic. In quiet or moderate rooms, most tools are close enough that workflow and privacy matter more than raw noise handling.

Is it legal to record an in-person meeting?

It depends on your jurisdiction and who’s in the room. Many US states use one-party consent (only you need to agree); others, and much of Europe, require all-party consent. In the EU, the GDPR treats a recording as processing personal data, so you need a lawful basis (usually consent). Wherever you are, the safe practice is the same: tell everyone before you start and get a clear yes — on the record. Devices that show a visible recording indicator make that disclosure obvious by design. (General information, not legal advice — consult a qualified lawyer for your jurisdiction.) is it legal to record a conversation · which states allow recording without consent

Are AI note takers private and GDPR-friendly?

Some are built for it; some aren’t — and you can’t tell from the marketing. The questions that separate them: Is your audio processed in the EU or shipped to a US cloud? Is the source audio deleted after the summary? Are your recordings used to train the vendor’s models? Jamie answers these well on the software side (EU servers, audio deleted after summary, ISO 27001). On the hardware side, Kuno’s answer is on-device transcription, EU hosting, and no training on your recordings. Whatever you choose, read the data-processing terms — don’t assume.

▶️ Privacy-first capture for the meetings bots can’t reach. Kuno is a recording device made in Germany: on-device transcription so audio never leaves the room, EU-hosted, never used to train AI, with a visible recording indicator and one-tap stop so consent is documented. No bot, no US cloud, no meeting link — just the in-person and field conversations captured cleanly. Get early access →

Common mistakes when choosing an in-person note taker

• Assuming “records on-device” means “private.” Many recorders upload to a cloud to transcribe. Check where the processing happens, not just the capture.
• Relying on a phone mic in a big room. Past a few feet, far-side speakers get muddy and the summary inherits the gaps.
• Forgetting offline reality. If the tool needs live Wi‑Fi and the client network is locked, your “live transcript” is a blank screen.
• Skipping consent. Recording first and disclosing later is both bad etiquette and, in all-party jurisdictions, illegal.
• Ignoring who owns the vendor. As the Limitless shutdown showed, a tool can change owners or exit your region overnight.

Troubleshooting: when in-person capture goes wrong

• Transcript is garbled at the far end of the table. Move the device to the center, reduce background noise, or switch to a multi-mic recorder; a single phone mic can’t cover a boardroom.
• The app won’t transcribe in the meeting. It likely needs a connection for live features — record offline and process afterward, or use a device that transcribes on-device.
• Speakers aren’t labeled. Diarization needs reasonably separated voices; people talking over each other defeats it. A central, multi-mic device helps.
• A recording stopped mid-meeting. Phone recorders can pause on an incoming call — a dedicated device avoids that failure mode.
• Legal/compliance flagged the tool. Check data residency and retention; an EU-hosted or on-device option is usually easier to clear.

Frequently asked questions

What is the best AI note taker for in-person meetings?

For teams, Fellow — it captures in-person audio bot-free and routes action items into the tools you already use. For dedicated hardware, Plaud leads on form factor. For EU privacy and data sovereignty, Jamie (EU-hosted software) and Kuno (a recording device made in Germany with on-device transcription) are the standouts.

Can an AI note taker join an in-person meeting?

No bot can join a physical room. Instead, bot-free tools capture audio through your device microphone or a dedicated recorder, then transcribe and summarize. That’s why in-person capture is a different problem from joining a Zoom call.

Do I need to tell people I’m recording?

Yes — disclose and get agreement before you start, regardless of local law. In all-party-consent regions and under the GDPR it’s required; everywhere else it’s best practice. A visible recording indicator makes the disclosure obvious.

Will an AI note taker work without Wi‑Fi?

A hardware recorder will capture audio offline; the AI summary then needs compute, in the cloud or on-device. Pure software apps that show a live transcript need a connection for the real-time part.

Is a dedicated device better than a phone app?

For larger rooms, client-facing meetings, or anywhere Wi‑Fi is unreliable, yes — better mic range, offline reliability, and a more professional presence. For quick informal chats, a phone app is fine.

Which option keeps my data in the EU?

On the software side, Jamie hosts on EU servers and deletes audio after the summary. On the hardware side, Kuno transcribes on-device with EU hosting and made-in-Germany hardware. Always confirm a vendor’s current data-processing terms.