Jamie AI Meeting Notetaker: No-Bot Speaker Identification for In-Room and Remote Speakers in Hybrid Meetings

Published October 27, 2026

October 27, 2026 · 11 min read

The scenario: six people around a conference table, three more on Microsoft Teams, and a leadership team that wants a clean transcript that actually says who committed to what. Can a no-bot AI meeting notetaker like Jamie tell the in-room speakers apart from the remote ones? The short answer is partially — Jamie captures both the computer microphone (your room) and the system audio (the Teams call) and runs speaker diarization across both, with its Speaker Memory feature labeling voices by name across future meetings. The honest caveat is that in-room audio shares one microphone, and that is a physics problem no bot-free tool fully solves. This buyer's guide walks through how Jamie handles it, where Granola and Microsoft Teams Rooms Intelligent Speakers differ, and when fully on-device apps like Basil AI are the better pick.

What Jamie AI actually is, in one paragraph

Jamie is a German-built, bot-free AI meeting assistant that runs as a native desktop app on macOS and Windows, plus an iOS app for in-person capture. According to Jamie's own documentation, it "doesn't join your call as a visible participant" and "captures audio in the background" on any platform — Zoom, Teams, Google Meet, or an in-person conversation around a table. A detailed Krisp vs Jamie comparison explains that during meetings Jamie records audio directly from your device — both the microphone and the speaker output — then produces a transcript with speaker labels and an AI summary.

How Jamie's speaker identification works in a hybrid meeting

Speaker identification (not just diarization) is Jamie's calling card. The Jamie product team explains that when a voice has not been identified before, Jamie asks you about that voice once, you type a name, and "Speaker Memory remembers that voice in your future meetings." On calendar meetings, Jamie also pulls names from the invite and attributes transcript lines automatically, so remote participants are usually tagged correctly before you even open the transcript.

In your hybrid leadership scenario, that means Jamie treats the three remote attendees on Teams the way any bot-free tool would — their voices come through the Teams audio stream and get labeled against their calendar names. The six in-room people share your laptop's microphone, so Jamie plays short audio clips after the meeting and asks you to name each new voice, as described in the bot-free notetaker guide. From the second meeting onward, those six voices are recognized automatically.

Where it works well

• Recurring leadership meetings with the same cast — Speaker Memory earns its keep after two or three sessions.
• Rooms where people take turns speaking without heavy crosstalk.
• Teams or Zoom calls where remote participants have named calendar invites.

Where it struggles

An independent product review on AI Tool Hunt lists "speaker identification may struggle with overlapping conversations" as a known limitation, and notes there are no live real-time captions. A Noota review of Jamie is more blunt: because Jamie is bot-free, "it may not automatically know who is speaking in the same way a meeting-platform-native assistant can."

The physics problem every bot-free notetaker runs into

If you want to understand why speaker identification is hard for every tool in this category, it helps to read a Granola post-mortem. In an excellent write-up on Granola's two-bucket labeling, the author explains that Granola's desktop app separates your microphone from system audio and labels them "Me" and "Them" — that is two buckets, not per-person labels. Splitting the four voices inside "Them" is "a genuinely hard signal-processing problem."

By contrast, when a notetaker bot joins as a participant, "Zoom or Meet or Teams hands it a separate audio stream for every person in the room" — attribution becomes bookkeeping, not inference. Granola's own speaker-tags documentation confirms that "if several people speak through the same shared device, Granola cannot distinguish the individuals using it," and that overlapping speech may not be assigned correctly. Jamie uses a different model (Speaker Memory with voiceprints), but the raw audio problem is the same.

Jamie vs Granola vs Teams Rooms vs on-device: a comparison table

Here is how the realistic options line up for a six-plus-three hybrid leadership meeting.

Tool	Meeting bot?	In-room speaker ID	Remote speaker ID	Where audio is processed	Audio retained?	Price (Oct 2026)
Jamie AI	No (bot-free)	Yes — ask-once Speaker Memory	Yes — from Teams/Zoom calendar names	EU servers (Frankfurt, Germany)	Deleted after transcription	Free tier; from €24/mo
Granola (desktop)	No	No — only "Me" / "Them"	Zoom names on macOS, Meet names on Mac + Windows	Cloud	Yes	From $18/user/mo
Granola (iPhone)	No	Yes — iOS mic diarization	N/A (in-person use)	Cloud	Yes	Same plan
Teams Rooms Intelligent Speakers	N/A (platform-native)	Yes — 7-mic array + voice profiles for up to 10 in-room people	Yes — platform-native	Microsoft Cloud	Yes	Teams Rooms Pro license + hardware
Basil AI (on-device)	No	Yes — Apple on-device diarization	Via system audio capture	100% on-device (iPhone / Mac)	You own it, delete anytime	Free + Pro tier

Granola in the same scenario: what you actually get

If your team is weighing Granola against Jamie, the honest comparison is this: on Mac and Windows, Granola's transcript will show "Me" (whoever is at the keyboard) and "Them" (everyone else, lumped together). Granola's transcription docs confirm that "the Granola mobile app (iPhone and Android) can also recognize different speakers during face-to-face meetings," but if you capture a hybrid room on a laptop, you get the two-bucket experience.

Granola's workaround guide for in-person meetings — summarized by Circleback — suggests placing the laptop centrally and even having everyone in the room "join a Zoom or Teams call from their own devices while sitting together" to give Granola per-person audio streams. That is a creative hack but it defeats the whole point of a bot-free experience, and it introduces feedback risk unless everyone wears headphones.

The hardware route: Microsoft Teams Rooms Intelligent Speakers

If you are already standardizing on Microsoft Teams for your leadership meetings, the official route to clean in-room speaker labels is Teams Rooms Intelligent Speakers. Microsoft's documentation explains that "in-room participants can be identified in live transcription, and all participants can easily see who's saying what during the meeting." According to the Microsoft Teams blog, the hardware uses a 7-microphone array to identify voices of up to ten people in meeting rooms.

The catch is operational. Each in-room attendee has to enroll a one-time voice profile, there must be no more than 20 people on the invite, the device needs to sit at least 8 inches from walls and laptops, and it requires a Teams Rooms Pro license. People who do not set up a profile — including anyone outside your tenant — show up as "Speaker 1" or "Speaker 2." For a mature enterprise this is fine; for a lean leadership team that occasionally has outside guests, it is a lot of ceremony.

Jamie's privacy posture, honestly stated

Jamie's privacy pitch is strong but it is not on-device. The Jamie security page and the detailed GDPR note-takers guide describe the architecture: AES-256 encryption at rest, TLS 1.2 in transit, EU-only processing on servers in Frankfurt, ISO 27001 certification, DORA readiness, meeting audio deleted after transcription, and no model training on your data. A July 2026 Neralvo review adds a sensible buying checklist: verify EU-residency contractually, confirm audio-deletion consequences, and understand transcript retention — strong privacy software doesn't mean regulators automatically treat it as sufficient.

For many European teams this clears the bar comfortably. For attorney-client privileged conversations, HIPAA-sensitive clinical reviews, or FINRA-regulated financial discussions, "audio leaves the room" is sometimes a conversation that is easier to never have. That is where on-device tools fit.

How Basil AI solves this: on-device transcription with no network at all

Basil AI is a privacy-first alternative built on Apple's on-device speech frameworks. As detailed by Apple at WWDC 2023, on-device speech recognition means the audio never has to leave the user's device, and "respecting the user's privacy is paramount." An October 2026 developer guide explains that Apple's new SpeechAnalyzer framework (shipping in iOS 26) is on-device only — no server fallback — with no duration cap and better distant-mic handling for meetings. That is the engine behind Basil.

In your six-in-the-room, three-on-Teams scenario, Basil AI captures microphone audio for the in-room voices and (on Mac) system audio from the Teams call, runs transcription and speaker diarization locally on the Apple Neural Engine, and never uploads the raw audio anywhere. If your Wi-Fi goes down mid-meeting, nothing changes. If your compliance team asks "where does the audio go?" the answer is: it stays on the device.

Trade-off: on-device apps can't call out to a huge cloud language model to generate a 20-page analysis of your meeting. But for the actual job — accurate transcript, speaker labels, action items, summary — the Apple-powered on-device stack is now competitive. For more on how that pipeline works end to end, read our deep dive on Apple Intelligence and private AI meeting notes.

Decision guide: which tool for which hybrid meeting

• Recurring internal leadership meetings, same cast every week: Jamie wins on in-room speaker labels because Speaker Memory pays off quickly. The data lives in EU cloud; make sure that's acceptable.
• Enterprise standardized on Microsoft Teams with budget for hardware: Teams Rooms Intelligent Speakers is the "official" answer, with per-person labels for up to 10 in-room attendees.
• Mac-first team that likes typing notes: Granola, but accept "Me" and "Them" labels on the desktop app for mixed rooms.
• Attorney-client, HIPAA-sensitive, or FINRA-regulated conversations: an on-device app like Basil AI — the audio never leaves the device, which materially shrinks your incident response surface. See our write-up on AI meeting notes for financial compliance officers.
• Hybrid team that mixes in-person workshops with remote 1:1s: on-device iPhone capture for the room + system-audio capture for Teams calls. Our comparison of Granola alternatives that are both bot-free and cloud-free walks through the architecture.

What this means for a leadership team buying today

If the question that led you here was "which no-bot AI notetaker can tell six in-room leaders apart from three remote participants on Teams," the honest answer stack looks like this:

• Yes, Jamie can do it — using Speaker Memory, after a short naming step on the first meeting — and it is a solid pick if EU cloud processing is acceptable to your compliance group.
• Granola can't do per-person in-room labels on desktop — it will show "Me" and "Them." Its iPhone app does better for in-person capture.
• Teams Rooms Intelligent Speakers is the gold standard for Microsoft shops that want per-person in-room labels and are willing to buy hardware and enroll voice profiles.
• Basil AI (on-device) is the pick when the audio cannot leave the room at all — regulated conversations, confidential strategy sessions, legal matters.

If you want to see a side-by-side breakdown of Jamie, Granola, Otter, Fireflies and Basil AI across pricing, platforms and privacy, our AI meeting notetaker comparison guide is the companion read.

Try Basil AI for private hybrid meetings

Basil AI is a fully on-device AI meeting notetaker for iPhone and Mac — real-time transcription, speaker diarization, 8-hour continuous recording, Apple Notes export, and zero cloud upload. If your leadership meetings include anything you wouldn't want sitting on a vendor's EU server after the call, Basil is designed to make that question irrelevant.