DEV Community: Julia Nebova

Why compliance teams are banning AI meeting assistants: A guide to hidden privacy risks in 2026

Julia Nebova — Fri, 23 Oct 2026 11:11:06 +0000

1. The EU AI Act Has Turned Meeting Privacy Into a Legal Deadline

€15 million or 3% of worldwide revenue. That's the exposure the EU AI Act created for organizations whose meeting assistants record calls without demonstrable disclosure — and it became enforceable on August 2, 2026.

The tools aren't the problem. One design choice is: a bot that joins the call as a participant and sends the audio to a third-party server. Compliance teams are banning that architecture, not AI itself.

2. Three Capture Models — and Why the Difference Is Legal, Not Cosmetic

A meeting bot does not just listen. It receives the same audio stream the platform sends to human attendees. Then it sends that stream to infrastructure the meeting host does not control.

Not all assistants work this way.

Capture model	How it records	Joins as participant?	Where audio usually goes	Examples
Bot-based	Virtual attendee joins the call	Yes	Vendor cloud; often long-term storage and model training	Otter.ai, Fireflies.ai, Fathom
Browser extension	Reads the active tab or live captions	No	Often still a vendor cloud	Tactiq, Bluedot
Device-level	Captures microphone + system audio on the user's machine	No	Depends on vendor — local or cloud	CatchMeet,Granola, Krisp

"Botless" does not mean private. A browser extension can capture locally and still upload everything to a cloud with unlimited retention and training rights. Privacy depends on three questions, not on whether a tile appears:

Where does the audio go after capture?
How long does it stay there?
What is the vendor allowed to do with it?

A tool with a visible bot and strict zero-retention can be safer than an invisible extension that trains models on your client calls.

3. The Legal Cascade: What Breaks When a Bot Is in the Room

If you're not a lawyer, here's the headline: a bot in a call creates at least three independent legal problems. Winning on one does not protect you from the others. Each can cost money on its own.

Risk	What triggers it	Who can be liable
Consent / wiretap	Bot records participants who did not agree	Vendor and deploying organization
Biometric privacy (BIPA)	Speaker identification creates voiceprints	Vendor and employer
Privilege waiver	Third party receives confidential communication	The organization claiming privilege
Discovery expansion	Permanent, searchable transcripts	The organization holding them
GDPR retention	Transcripts kept without a defined limit	Controller (your organization)

Consent and wiretap

US federal law needs one party's consent. Thirteen states need everyone's: California, Connecticut, Delaware, Florida, Illinois, Maryland, Massachusetts, Michigan, Montana, Nevada, New Hampshire, Pennsylvania, and Washington. A bot joins because the host asked — but every other participant never agreed.

That's the theory in Brewer v. Otter.ai (four consolidated class actions, Aug–Sep 2025). On August 13, 2026, the court denied the motion to dismiss on core wiretap, California privacy, and Illinois biometric claims — finding it plausible Otter used conversation data for its own commercial purposes.

"The host consented" is no longer a defence that ends the case.

Biometric privacy (BIPA)

Speaker identification needs voiceprints — biometric data under Illinois' BIPA, which requires written notice, written consent, and a published retention policy first.

Cruz v. Fireflies.AI Corp. (filed Dec 18, 2025): an Illinois resident joined a normal meeting, never signed up for any AI tool, never consented. A bot joined at someone else's request, recorded the call, named speakers, produced attributed transcripts.

Why employers care: BIPA lets multiple parties be liable for the same collection. If any participant was in Illinois when recorded, BIPA applies regardless of where you're based.

Privilege and the third-party problem

Privilege depends on confidentiality. Put a vendor in the room, and opposing counsel has an argument. You do not need a human at the vendor to have listened. You need the vendor to have access under its terms.

4. Why "Just Configure Zoom or Teams" Falls Short

Platform controls have improved. But they all share the same limit: they govern admission to the room.

Microsoft Teams added "Unverified" bot labels (June 2026) and auto-blocking of detected external bots (August 2026). Zoom requires host/co-host to start AI Companion. Google Meet gates "Take notes for me" by org unit.

All three share a blind spot: a device-level tool never asks for admission. It captures system audio on the user's machine. No lobby entry, no bot signal, no tenant policy applies. Microsoft's own documentation concedes this — auto-blocking depends on detecting a participant as a bot and does not prevent local recording or screen capture.

The second gap: bans push usage into shadow channels. Block the approved tool and employees connect a personal Otter account to their work calendar. That bot arrives through an identity the admin doesn't manage. Prohibition without an approved alternative produces concealment, not compliance.

5. The Architectural Shift: Device-Level Capture

The question that matters is not "AI or no AI." It is: does the assistant become a participant in the conversation?

A bot-based assistant enters the room — triggering the consent, privilege, and biometric problems above. A device-level assistant does not enter the room. It captures audio on the user's device and adds no participant. The platform sees nothing join.

That difference is legal, not cosmetic. If there is no third party in the conversation, the wiretap argument and the privilege-waiver argument weaken by design.

CatchMeet is one example of this architecture: it runs on the user's computer, captures system audio from Zoom/Meet/Teams/Slack/Discord, adds no participant, stores data locally, and routes any processing through providers under zero-retention agreements.

The honest caveat: device-level is not a privacy guarantee. A botless tool can still upload everything to a cloud with unlimited retention and training rights — at which point the core risk returns through a different door. And device-level capture does not eliminate consent law: in an all-party-consent state, recording without disclosure is still a violation regardless of how the audio was captured. What device-level removes is the third-party-in-the-room problem — the specific trigger for the privilege, eavesdropping, and BIPA arguments above.

6. Compliance Checklist: What to Do This Week

#	Action	Why	Owner
1	Inventory every AI tool, including personal accounts connected to work calendars	A tenant block is useless if a personal bot is already syncing	IT / Security
2	Classify each tool by capture model (bot / extension / device-level)	Architecture decides legal exposure	Security / Legal
3	Check: does audio leave the endpoint, where is it stored, for how long, training rights?	"Botless" is not a privacy claim	Procurement
4	Harden platform controls (Teams, Zoom, Meet)	Necessary baseline — covers admission only	IT
5	Review endpoint applications separately from meeting settings	Device-level tools bypass bot policies	IT / Endpoint
6	Set retention periods by meeting category; require documented deletion	GDPR storage limitation; AI Act proof requirements	Legal / DPO
7	For legal, HR, M&A, board calls: allow only architectures that add no third party	Privilege and consent risk highest here	Legal
8	Provide an approved alternative employees actually want to use	The only real fix for shadow AI	Security / Ops

Steps 1–5 are inventory. Steps 6–8 are the ones most organizations skip — and the ones that decide whether the policy survives contact with employees.

7. FAQ

Are AI meeting assistants illegal under the EU AI Act?
No. Article 50 requires transparency — disclosure and labelling — not a ban. Fines reach €15M or 3% of turnover for non-compliance. The burden falls on the deploying organization.

What makes bot-based assistants riskier?
They introduce a third party into the conversation. That single feature triggers consent, privilege, and biometric rules simultaneously. See Section 3.

Does "botless" mean private?
No. It describes capture method, not data flow. Three questions decide privacy: where audio goes, how long it stays, and what the vendor can do with it. See Section 2.

Which US states require all-party consent?
Thirteen: California, Connecticut, Delaware, Florida, Illinois, Maryland, Massachusetts, Michigan, Montana, Nevada, New Hampshire, Pennsylvania, and Washington.

Can a bot waive attorney-client privilege?
It can create the argument. A vendor with contractual access to transcripts is a third party. The Brewer v. Otter ruling makes this argument credible.

What is the BIPA risk for employers?
BIPA allows multiple parties to be liable for the same biometric collection. An employer that selects or benefits from the tool can be sued alongside the vendor. See Section 3.

Can blocking external bots in Teams or Zoom solve this?
Partially. Those controls detect bots that try to join as participants. They don’t cover device-level capture, local recording, or personal accounts the admin doesn’t manage. See Section 4.

AI Meeting Assistants Market in Italy: Landscape and Top 5 Tools for Italian Teams (2026)

Julia Nebova — Thu, 22 Oct 2026 21:12:59 +0000

1. Introduction: A Growing Market, an Underserved Meeting

Italy's AI market hit €1.8 billion in 2025, up 50% year over year, according to the Politecnico di Milano's AI Observatory. 46% of that market came from generative AI or hybrid projects. 71% of large Italian companies had launched at least one AI project, against just 8% of SMEs. And by August 2026, AI assistants reached 18.9 million unique visitors in Italy, double the roughly 9 million of January 2025 (Sensemakers on Comscore data).

Only 23% of Italian workers see their meetings as genuinely useful, and 56% still don't use AI to manage them (ASUS Business survey). You can have a €1.8 billion market and still have a meeting culture that runs on manual notes and post-call memory.

The gap isn't transcription. Commodity tools have handled that for two years. The gap is trust, language, and context: GDPR constraints, Italian regional accents, and client conversations that an English-first model can't reliably parse without a cleanup pass.

This guide maps the Italian market, sets out the criteria that matter when choosing an AI meeting assistant, and ranks the top 5 tools for 2026 by fit for the Italian operational reality, not by feature count.

Disclosure: CatchMeet is our product. We scored it against the same criteria as everyone else, listed its limits, and linked our sources so you can check them. The best test is still a 30-minute call in Italian on your own meetings.

2. The AI Meeting Assistants Market in Italy: Size, Drivers and Players

How big is the AI meeting assistants market in Italy?

No analyst publishes an Italy-only figure yet, so the honest answer is a triangulation.

Globally, analysts put the market at $2.5–2.7 billion in 2024: $2.53 billion according to The Business Research Company and $2.68 billion according to Market.us, which expects it to reach $24.6 billion by 2034 at a 24.8% CAGR. North America alone holds more than 40% of the market, so Europe is still catching up. Technavio expects the market to add $3.46 billion between 2024 and 2029 at a 28.2% CAGR, with remote and hybrid work as the main driver.

For Italy, the best proxies are the size of the AI market (€1.8 billion in 2025, see above) and the number of people who work remotely and live in video calls. As an illustration, not a forecast: if one in ten of Italy's 3.575 million smart workers paid €15 a month for a meeting assistant, that alone would be worth about €64 million a year.

What drives demand in Italy

Hybrid work is stable, not fading. Italy counted 3.575 million smart workers in 2025, up 0.6%. Smart-working programs exist in 95% of large companies and 67% of public administrations, but only 45% of SMEs, down 8 points in a year (Il Giorno, on Politecnico di Milano data).
AI assistants are becoming a habit. 40% of Italians online now use an AI assistant, and time spent rose to 80 minutes a month in the first half of 2026, from 44 a year earlier (Spot and Web, on Sensemakers data).
SMEs are the untapped segment. AI projects are far more common in large companies than in SMEs (see the introduction), so most of the growth headroom sits in smaller firms.

Regulation shapes how Italians buy

In Italy an AI note-taker sits under four layers of rules at once: GDPR, the Workers' Statute (Statuto dei Lavoratori) on remote monitoring of employees, the EU AI Act, whose transparency obligations apply from 2 August 2026, and Italy's own AI law, Legge 132/2025 (Filodiritto). In practice, that means notice before recording, a defined purpose, and a rule that transcripts made for minutes don't become material for judging individual performance (Tom's Hardware Italia). Vendors that make these answers easy have an advantage with Italian procurement teams.

Who plays: four groups

Global standalone tools: Fireflies, Otter, tl;dv, Granola, Fathom. Broad features, mostly English-first.
Built-in platform AI: Gemini in Google Meet, Copilot in Teams, Zoom AI Companion. Already paid for by many companies, limited to their own platform.
European privacy-first tools: Jamie, Noota, Leexi, HappyScribe, selling EU data residency as the core promise; Jamie and HappyScribe also record without a bot (Noota).
Device-level and on-device capture: ASUS AI ExpertMeet, Wispr's Notetaker, CatchMeet. No bot in the call, with audio captured on the user's machine.

Trends to watch in 2026

The bot is leaving the call. Granola, tl;dv's new desktop app and on-device tools all move capture off the participant list.
Built-in tools are catching up on Italian. Gemini and Copilot both handle Italian now, which raises the bar for standalone vendors on everything except cross-platform use and real-time help.
Compliance becomes a feature. With the AI Act's transparency rules in force, consent prompts, data location and retention controls move from the legal annex to the sales pitch.
From notes to in-call help. The next step after summaries is assistance during the conversation: answers, fact-checks, context.

The building blocks exist across these four groups. What's still rare is the combination: Italian treated as a first-class language, capture without a visible bot, data under European control, and help during the conversation rather than after it.

Positions reflect this article's assessment of each tool, October 2026.

That's the gap. Now let's define what a good answer looks like.

3. How to Choose an AI Meeting Assistant: Five Criteria

Before the ranking, we need to be explicit about the evaluation logic. Five criteria separate a tolerable tool from the right one. Each tool is scored 1 to 5 on each of them in section 5.

Italian-language performance

Not "Italian support." Whether the engine handles regional accents, code-switching and technical vocabulary without a post-editing pass. This is the criterion most vendors quietly fail, and the only public benchmark so far is vendor-run.

tl;dv Italian meeting transcription benchmark · run by tl;dv, so read it as one data point.

Capture architecture: bot or device-level

A bot in the participant list is visible to your client, and it changes what people say. Device-level capture records from your computer instead. One caveat matters in Italy: no bot does not mean no consent. Italian courts allow a participant to record a conversation they take part in, but processing that recording for business purposes still falls under GDPR (Il Sole 24 Ore). Bot-free capture removes the distraction, not the duty to tell people.

EU data residency and GDPR

For Italian legal, financial, healthcare and public-sector teams, this is a procurement prerequisite, not a differentiator. The question is not whether a vendor says "GDPR-compliant" but where audio, transcripts and AI processing physically happen, and whether you can get a DPA. With most of the EU AI Act applicable from 2 August 2026, procurement teams are asking harder questions (Filodiritto).

Real-time help vs post-meeting summary

Most tools summarize after the call. Higher value: helping during the conversation, with answers and fact-checks while you can still use them.

Cross-meeting memory and platform coverage

A tool that treats each meeting as an isolated event can't see the trajectory of a client relationship. And a tool that runs well on only one OS excludes part of your team.

4. Top 5 AI Meeting Assistants for Italian Teams (2026)

The order below follows the scores in section 5.

1. CatchMeet — Best Overall for Italian Teams That Need Privacy and Context

CatchMeet is a bot-free, real-time AI meeting assistant built around device-level capture. It records audio directly from the operating system (mic plus system output), so it works across Zoom, Google Meet and Teams without showing up as a participant.

In our testing on a 40-minute client call with a Milan-based consultancy, the absence of a visible bot kept attention on the conversation rather than on the recorder. (No bot doesn't remove the need to tell people you're taking AI notes; see section 3.)

What works:

No bot. Audio is captured locally, so the participant list shows only people.
Real-time answers and fact-checking. It surfaces answers from your docs and connected sources and checks claims during the meeting (CatchMeet on Product Hunt). That's a different capability from post-call notes.
Cross-meeting memory. Context builds up across meetings, so you stop re-briefing the tool every call.
A usable free tier. 120 minutes per month with real-time answers, live transcription and summaries. Paid tiers go from 12 to 40 hours of live coverage per month, and silence doesn't consume minutes.
Italian as a working language. Live transcription and answers work in Italian. We haven't published an independent Italian accuracy benchmark yet, so run a test call on your own accents and vocabulary.

Where it's not perfect: there's no native mobile app yet, so capture is tied to desktop. Integrations on the free tier are limited. Cross-meeting memory needs a few meetings of history before it pays off. And as with every tool here, ask for the DPA and where transcription and AI processing run before you sign.

Best for: Italian consultancies, sales teams, legal and financial firms, any team where the meeting is a confidential conversation with a client rather than a demo.

2. tl;dv — Strong Italian Benchmarks, Post-Call Value

tl;dv has made a credible bet on European language coverage. Its own Italian benchmark scored 181/200, with particular strength in transcription accuracy. Pro costs $18 per seat per month on annual billing, and the free plan records without limit but caps AI notes (Laxis tl;dv review).

By default it joins as a bot. Since 2026, a desktop app for Mac and Windows can record in the background without one, which narrows the gap with bot-free tools. Two limits remain: every recording is capped at three hours (Costbench), and the value is post-meeting rather than in-meeting assistance.

If Italian transcription accuracy on a budget is your priority, tl;dv is the strongest number on paper. Just remember the benchmark is vendor-run, and test it on a long call before you commit.

3. Granola — Premium Bot-Free Notes, Without Real-Time Help

Granola delivers a well-designed bot-free experience. It captures system audio locally, summaries are polished, and shared folders let a team query past calls with AI chat. Pricing is a free Basic plan, Business at $14 per user per month and Enterprise at $35, billed monthly only (Granola).

The limitations are about timing and language rather than platform. Granola now runs on Mac, Windows and iOS, but its help comes after the call, not during it, and it publishes no Italian accuracy data. It holds SOC 2 Type 2 and offers a DPA on request; confirm data location for regulated work.

Granola is a strong choice if your team wants beautiful notes and shared history, and doesn't need answers while the client is still talking.

4. Fireflies.ai — The Ecosystem Play, With Caveats

Fireflies has the broadest integration surface in this ranking. Pricing starts at $10 per seat per month on annual billing for Pro and $19 for Business (Castmagic).

The problem is the bot. It joins visibly, as "Fred" (ScreenApp), and for many client-facing Italian teams that alone is a non-starter. Custom data residency sits on the Enterprise plan. Italian transcription is usable but not excellent (142/200 in tl;dv's benchmark), and advanced features consume AI credits, so cost is hard to predict at scale.

If you're looking for a Fireflies alternative without a bot, CatchMeet, Granola and tl;dv's desktop app are the options above. Fireflies still wins if ecosystem breadth matters more to you than capture privacy. That's a legitimate trade-off, as long as you make it knowingly.

5. Otter.ai — The English Standard, Not the Italian One

Otter is one of the most mature real-time transcription engines for English. The free tier gives 300 minutes per month, and Pro is $8.33 per user per month on annual billing (Sonix).

For Italian, it's not a viable production tool: Italian isn't among its six transcription languages (HappyScribe). The bot is visible, and the US class action over recording without consent (Laxis) makes it a hard sell to an Italian compliance team. If you're running a legal practice in Rome, Otter isn't a serious candidate.

If your team works in English, Otter is fine. If you need an Otter alternative for meetings in Italian, look higher up this list.

What About Built-In Tools: Gemini, Copilot, Zoom?

Many Italian companies already pay for an AI note-taker inside their video platform, so it's fair to ask whether they need another one.

Gemini in Google Meet. "Take notes for me" supports Italian, notifies every participant, and saves notes to Google Docs (9to5Google). It only works in Meet, handles one language per meeting, and scored 118/200 in tl;dv's Italian benchmark.
Copilot in Microsoft Teams. Intelligent meeting recap works in Italian, including translated recaps for multilingual meetings, but it needs a Teams Premium or Microsoft 365 Copilot license and transcription switched on (Microsoft Learn). Zoom AI Companion offers similar summaries inside Zoom (Neowin).

Built-in tools are the easiest choice if your whole team lives in one platform and you only need a recap. They fall short when your clients use a different platform than you do, or when you need help during the call rather than notes after it.

5. Side-by-Side Comparison

Table A. Pricing (publicly listed, October 2026)

Tool	Free tier	Paid entry	Team / Business	Notes
CatchMeet	120 min/mo, real-time answers	Pro: 20 h live coverage/mo (tiers range 12–40 h)	Higher tiers with Jira, Linear, Confluence	Silence doesn't consume minutes
tl;dv	Unlimited recording, ~10 AI notes	Pro $18/seat/mo (annual)	Business $29/seat/mo (annual)	3-hour cap per recording
Granola	Basic, limited history	Business $14/user/mo (monthly only)	Enterprise $35/user/mo	Windows still beta in some reviews
Fireflies	400 min storage, 20 AI credits	Pro $10/seat/mo (annual)	Business $19/seat/mo (annual)	Advanced features consume AI credits
Otter	300 min/mo	Pro $8.33/user/mo (annual)	Business $19.99/user/mo (annual)	Italian not supported

Prices change often; check the official pages before you buy: Otter, Fireflies, tl;dv, Granola.

Table B. Criteria scoring (1–5)

Criterion	CatchMeet	tl;dv	Granola	Fireflies	Otter
Italian production quality	4*	4	3	3	1 (not supported)
Bot-free capture	5	3 (desktop app)	5	1	1
EU data control	4*	4	2	3	1
Real-time assistance	5	1	1	2	2
Cross-meeting memory	5	3	3	3	3
Cross-platform	5	5	4	5	5
Total	28	20	18	17	13

Scores are our judgment against the criteria in section 3, based on vendor documentation, the sources linked above and hands-on use where available. * Our own assessment of our product: there is no independent Italian benchmark for CatchMeet yet, so test it yourself.

6. Quick Decision Guide

Start with the first line that describes your team:

Regulated sector (healthcare, public sector, banking)? → Before comparing features, get the DPA, the processing location and a consent notice for participants. Then shortlist only tools that pass.
Whole team on one platform, and a recap is enough? → Use the built-in tool: Gemini in Meet, Copilot in Teams or Zoom AI Companion.
Confidential client calls across different platforms (law, finance, consulting)? → Bot-free capture. CatchMeet if you need answers during the call; Granola if post-call notes are enough.
Mostly Italian, tight budget, a bot is acceptable? → tl;dv. Test a long call first, and remember the three-hour cap.
Sales team that lives in the CRM, bot acceptable? → Fireflies.
Meetings mostly in English? → Otter.

The common thread: for Italian teams in regulated or client-facing work, bot-free capture and EU-aligned data control are the entry ticket, and consent still applies either way.

7. FAQ

How big is the AI meeting assistants market in Italy?
No analyst publishes an Italy-only figure. Analysts put the global market at $2.5–2.7 billion in 2024, growing roughly 25–28% a year. In Italy, the main demand signals are 3.575 million smart workers, AI projects in 71% of large companies, and 18.9 million monthly visitors to AI assistants.

What is the best AI meeting assistant for Italian teams in 2026?
It depends on the meeting. For confidential client calls where you want help during the conversation, our pick is CatchMeet (our own product, see the disclosure above). For accurate Italian transcription on a budget, tl;dv. For teams that live entirely in Google Meet, Gemini's built-in notes.

What are the top 5 AI meeting assistants for Italian teams?
In order: CatchMeet, tl;dv, Granola, Fireflies.ai and Otter.ai, ranked by Italian-language performance, capture architecture, EU data control, real-time assistance, memory and platform coverage.

What are the best alternatives to Fireflies and Otter for meetings in Italian?
If the bot is the problem, look at bot-free tools: CatchMeet, Granola, or tl;dv's desktop app. If Italian accuracy is the problem, tl;dv has the strongest published Italian score, and Otter doesn't support Italian at all.

Are AI meeting assistants GDPR-compliant in Italy?
Compliance depends on data architecture, not a badge. Fireflies offers custom data residency on its Enterprise plan; Otter faces a US class action over recording without consent; CatchMeet and Granola capture audio on the device. For any tool, confirm where transcripts and AI processing run, and sign the DPA before rollout.

Do AI meeting assistants handle Italian regional accents and technical terminology?
Most global tools underperform. Otter doesn't support Italian. In tl;dv's own benchmark, tl;dv scored 181/200, Fireflies 142, Gemini 118 and Fathom 114. Run a 30-minute test call with your own accents and vocabulary before committing.

What is the difference between bot-based and bot-free meeting assistants?
A bot-based tool (Fireflies, Otter, tl;dv by default) joins as a visible participant. A bot-free tool (CatchMeet, Granola, tl;dv's desktop app) captures audio on your device. Built-in tools such as Gemini in Meet run inside the platform and notify participants.

Is it legal to record meetings with AI in Italy?
Italian courts allow a participant to record a conversation they take part in, but processing that recording for business purposes falls under GDPR and needs a legal basis. Best practice: tell participants at the start, whatever tool you use. For employee monitoring and sector-specific rules, consult Italian counsel.

8. Conclusion: Where the Italian Market Is Heading

The Italian AI assistant market has crossed the adoption threshold. The meeting-assistant category is still catching up. Most global leaders were built around English-first models, bot-based capture and post-meeting summaries, and those defaults show their limits under Italian conditions.

The direction is clear: assistants that stay out of the participant list, keep data under European control, and help during the conversation rather than after it. That's the bet we made with CatchMeet. Whichever tool you shortlist, judge it on a real Italian call, ask where your data goes, and tell your participants it's there.

How to Block AI Notetakers From Your Meetings (Zoom, Teams, Meet)

Julia Nebova — Tue, 13 Oct 2026 07:06:26 +0000

You are ten minutes into a sensitive call when a name nobody recognizes appears in the participant list: Fireflies.ai Notetaker. Nobody on your side invited it. The conversation slows, and someone asks the question you hoped to avoid: who is this, and where is the recording going?

You can stop most of these bots, but not with one switch. Three layers work together: stop the invitation at its source, make the platform check who may join, and put every unknown joiner in a lobby where a person decides. A fourth step, removing a bot that got in, is the backstop.

One limit comes first. Blocking bots does not stop a person from recording with a desktop app on their own computer. Otter's desktop app and Chrome extension work without a bot in the meeting, and Fireflies, Fathom and tl;dv also offer or advertise bot-free modes. No meeting platform setting can see those. For them you need a recording policy, not a block list.

The short answer: what you can block and what you cannot

The fastest fix, about ten minutes:

Hold unknown joiners in the lobby. Teams detects external bots and holds them in the lobby by default. In Zoom, turn on the waiting room. In Google Meet, set the meeting access type to Trusted or Restricted. In GoTo, lock the meeting.
Require identification. Require sign-in in Zoom, or block external email domains in Webex Control Hub.
Switch off auto-join at the source. Ask the person whose calendar invited the bot to turn off its auto-join. In Fireflies that is the Auto-Join setting, and its Bar Rules can exclude meetings by title or keyword.

Then give people a bot-free alternative. If you only block, some people will move to a recorder you cannot see. An approved bot-free tool keeps the notes without a tile in front of clients. CatchMeet is one: it runs on the user's computer, adds nothing to the participant list, and has a free plan.

Sections below give the exact settings per platform, the admin checklist, and the reasons blocks sometimes fail.

How notetaker bots get into a meeting

A bot reaches your meeting by one of a few routes, and each route is stopped at a different checkpoint.

Route	How it works	Where to stop it
Calendar auto-join	The bot tries to enter at the scheduled start time of any event on a connected calendar. If a lobby is on, a host must admit it (Fireflies guide)	The notetaker's own auto-join setting; the lobby
Invited address or link	Someone invites the bot's address (for Fireflies, `fred@fireflies.ai`) or pastes the meeting link into the vendor's dashboard	Sign-in and domain rules; the lobby
Marketplace integration	Official apps use the Zoom Meeting SDK or Zoom RTMS, which hosts can allow or block (Zoom)	Zoom's Meeting SDK Apps block list
Automation in the web client	Unofficial tools drive Zoom's native or Web App. Zoom says Web App bot protection is on for all accounts and cannot be turned off (Zoom)	Built in; sign-in requirement

These bots look like ordinary participants, and that is on purpose. Fireflies says its bot "is always visible in the participant list" under the name Fireflies.ai Notetaker, and Read advertises that "everyone in the meeting knows Read is there." The risk is not secrecy. It is that nobody decided to admit it.

Removing a bot once does not solve the problem. A bot that joins from calendar auto-join returns the next time the event recurs, because the invitation, not the room, is what brings it. That is why the blocks in this guide start before the meeting.

Block at the earliest checkpoint you control. The later ones catch what slips through. Sources are listed at the end.

Host settings by platform: Zoom, Teams, Meet, Webex and GoTo

Host settings work meeting by meeting and depend on someone remembering them. Use them as the first line of defense, and put the durable controls in the admin section. Where a vendor's page did not give a setting's exact path, the table says so instead of guessing.

Platform	Hold joiners in a lobby	Require identification	Bot-specific control	What the vendor page leaves open
Zoom	Waiting room; guests who are not signed in can be held for host review	Require sign-in to join; email-domain rules	Meeting SDK Apps block list; bot blocking dashboard; Web App bot protection (always on)	The block-list and dashboard articles returned no readable text when checked, so menu paths are not given
Microsoft Teams	Lobby; detected bots are held there even if the meeting lets others bypass it	Anonymous-join and lobby-bypass settings	Bot detection with a lobby label, on by default	Detection can miss bots and can flag people
Google Meet	Access type: Open, Trusted or Restricted	Restricted admits only invited guests	None found in Google's help pages	No bot detection or bot label found
Webex	Lobby with Let in, Remove and Move to Lobby; site-level auto-lock in Control Hub	Domain rules in Control Hub (admin)	None at host level	Lock and expel steps are not in the lobby guide; bots with no email address are not addressed
GoTo	Locking a meeting sends latecomers to a waiting room	None found	None found	Dismiss is documented for Classic (People pane); the current app's menu was not checked

Zoom

Zoom's waiting room article lists the controls that matter: require sign-in to join, restrict by email domain, use the waiting room, and manage which Marketplace apps may join. For a single meeting:

Turn on the waiting room under Settings > Meeting > Security, then choose Edit Options under Waiting Room Options.
Pick who waits. The option "Users who are not in your account and not part of your whitelisted domains" lets people from listed domains in and holds everyone else, and it is a waiting-room rule rather than a hard block (Zoom).
When a name you do not recognize appears, deny it from the waiting room. If it is already in, open the participant list, choose More next to it and remove it.

Microsoft Teams

Teams labels external bots it detects and holds them in the lobby "regardless of the meeting's lobby configuration," and an organizer must admit them explicitly (Microsoft Learn). As the organizer:

Limit who can admit people from the lobby to organizers and co-organizers, so a presenter cannot admit a bot by mistake. Microsoft recommends this.
Treat any flagged participant as a decision, not a formality. Teams shows a risk reminder before you admit one.
If a real guest is flagged, admit them and choose This is not a bot. That applies to that meeting only.

Google Meet

Meet gives hosts access types rather than a bot filter (Google Workspace help):

Set the access type to Trusted (people in your organization, plus external guests invited through a Calendar link) or Restricted (only guests invited through Calendar, plus people a host invites in the meeting).
Uncheck "Anyone with the meeting link can ask to join" for Trusted or Restricted meetings, so other guests are declined automatically.
Turn on Host management (Calendar event > Video call options, or Host controls during the meeting) so only hosts and co-hosts can eject participants and accept knocks. Turn on "Host must join before anyone else."

Webex

Webex's article on preventing AI bots covers admin-level domain rules only (see the admin section). For hosts, Webex's lobby guide describes the practical steps. When someone is waiting, a notification appears in the meeting controls, or in the Participants panel if it is open. The list is sorted into Internal, External and Unverified, so you can act on a whole group at once. Tick the people (or choose Select all) and click Let in or Remove. To send someone who is already in back to the lobby, right-click the name and select Move to Lobby. That guide does not describe locking a meeting or expelling someone after they joined, and the lobby feature is not available in Webex Meetings for Linux.

GoTo

In GoTo, an organizer can lock the meeting. People who try to join while it is locked wait in a waiting room, and everyone waiting connects automatically when you unlock. To remove someone who is already in, the organizer right-clicks the attendee's name in the People pane, selects Dismiss and confirms with Yes; the attendee leaves and is told they were excused. That page is for GoToMeeting Classic, so check the menu in your current app. GoTo's pages do not mention bots: lock the meeting once everyone expected has arrived, and dismiss anything unfamiliar.

Admin controls and the org-wide checklist

Host settings fail the day a host forgets. Admin controls take the decision away from the host, which is why they belong in every plan. What each platform offers differs a lot.

Zoom. Zoom lets admins allow or block specific Meeting SDK apps published on its Marketplace, and says unpublished SDK apps cannot join meetings outside the developer's account. Its bot blocking dashboard in the Meetings app shows which apps are attempting to join, joining or being blocked. Waiting-room defaults can be set for the whole account under Admin Center > Settings > Meeting > Security, or per group (Zoom).

Microsoft Teams. Bot handling is a meeting-policy setting under Teams admin center > Meeting policies > Meeting Join and Lobby > Manage external bots and their access to meetings. In PowerShell it is ExternalBotAccessMode. It can be set for the whole tenant or for targeted groups (Microsoft Learn).

Admin center label	PowerShell value	Behavior
Do not detect bots	`AllowAllBots`	Bots are not detected or marked and appear as ordinary external participants
When detected, require approval before joining	`RequireApprovalWhenDetected`	Default. Detected bots are marked and held in the lobby until explicitly admitted

Microsoft Message Center item MC1459141 (roadmap ID 566201) describes a further policy that blocks identified external bots automatically. The listing gives its default state as Off, with worldwide general availability starting in early October 2026 and finishing by late October. The listing names no PowerShell value, and Microsoft Learn's reference still lists only the two values above, so confirm the option in your tenant's Teams admin center before scripting it. Learn's text names Set-CsTeamsMeetingPolicy, but its examples use Set-CsTeamsEventsPolicy, so check the cmdlet too. Microsoft says some bots will go undetected and some people will be flagged.

Google Workspace. Admins set whether Host management is on by default (Google), and can gate Gemini's "Take notes for me" by organizational unit or group. Google's documentation lists no admin setting that filters third-party notetaker bots.

Webex. In Control Hub, go to Services > Meeting > Settings and, under Meetings, choose an option next to Internal. "Block external users with selected email domains" is the option that blocks bots by domain. Enable the Site Admin feature "Request attendee email address for all sites" so the block works as intended. The article is dated 14 April 2026, applies to the Webex Suite meeting platform, and says it replaces the Site Admin "Blocked email list" (Webex). It does not say what happens to bots that join with no email address.

Webex lobby and auto-lock. Site-wide lobby behavior sits in Control Hub under Services > Meeting > Sites: choose the site, open its settings, then Security under Common Settings. When a meeting is unlocked offers three choices for guests: join directly, wait in the lobby until the host admits them, or cannot join. Automatically lock meeting locks it 0, 5, 10, 15 or 20 minutes after the start (default 10), and applies only where hosts have not set their own preference. When a meeting is locked decides whether late arrivals wait in the lobby or cannot join at all.

The org-wide checklist

Inventory what is connected. List the notetaker apps people already use, and the calendar and OAuth apps your identity provider has consented to.
Move from block lists to allow lists where the platform offers it. A block list only covers names you already know. Zoom's Meeting SDK allow list and Webex's approved email domains are examples.
Set defaults that do not depend on hosts. Detect-and-hold in Teams, Host management on in Meet, sign-in required in Zoom.
Limit who can grant calendar and OAuth access. Auto-join starts with a user connecting a calendar. Restricting that grant removes the route at its source.
Publish a short internal policy. Use the template below.
Monitor and review. Check Zoom's bot blocking dashboard and Teams audit logs for unusual participants, and review settings every quarter, because vendors keep changing them.

AI notetaker and recording policy (template)

Approved tools: only [tool names] may record or transcribe company meetings.
Announce it: the host says at the start that an AI assistant is running and waits for objections.
Objection: if anyone objects, recording stops and notes are taken by hand.
External meetings: external participants are held in the lobby; only the host or a co-host admits them.
Unknown participants: anyone not recognized is denied admission and reported to [IT contact].
Calendar access: employees may not connect third-party notetakers to their calendars without approval.
Regulated content: calls covering health, legal or financial matters need [compliance owner] sign-off before any recording.
Storage: recordings and transcripts are stored in [location] and deleted after [period].
Review: IT reviews settings and the approved-tools list every quarter.

Why bots still join after you block them

If a bot keeps returning, one of four things is usually happening.

The invitation never stopped. A bot that follows calendar events keeps trying every time an event recurs. Removing it from the call fixes one meeting. The lasting fix is the owner's auto-join setting, or an admin limit on who can connect calendars.

The block does not match how the bot joins. A domain rule only sees the email domain of the joiner. It does nothing against a bot that joins from an account on a domain you allow, or one that is bot-free and never joins at all. Webex's article does not say how its domain block treats bots that join without an email address.

Detection has gaps. Microsoft says some external bots will go undetected and some people will be flagged as bots. Zoom says a real user flagged by its Web App protection can still sign in to join.

The tool is not a bot. Fireflies has a no-bot "Take Notes" mode in its desktop app, Fathom says you can capture notes "bot or no bot," tl;dv says "no bot required," and Otter's desktop app records without one. CatchMeet works the same way: it runs on the user's computer and adds no tile. No bot filter touches these.

Notetaker names and domains worth checking

Use this as a starting list for allow-and-block decisions, not as a finished block list. The address a bot shows can differ from its vendor's website domain, so confirm what appears in your own participant list before you block anything.

Tool	What to look for	Bot-free option exists
Fireflies	Name Fireflies.ai Notetaker; invite address `fred@fireflies.ai` (guide)	Yes: a no-bot "Take Notes" mode in the desktop app
Otter.ai	Domain `otter.ai`	Yes: desktop app and Chrome extension
Read.ai	Domain `read.ai`	Not stated on its home page
Fathom	Domain `fathom.ai` (`fathom.video` now redirects there)	Yes: "Now available bot-free"
tl;dv	Domain `tldv.io`	Yes: "No bot required"
CatchMeet	Nothing to block: it captures audio on the user's computer and adds no participant tile	Always: device-level capture on Mac and Windows

A starter block list. Only fireflies.ai has a bot address confirmed on the vendor's own page. The other four are the vendors' website domains, and a bot may join from a different address, so treat the list as a first pass and check it against your own participant lists.

# Starter list of notetaker domains
# Bot address confirmed on the vendor's own page
fireflies.ai     # invite address fred@fireflies.ai

# Vendor website domains; bot address not confirmed
otter.ai
read.ai
fathom.ai
tldv.io

Webex accepts a list like this under Block external users with selected email domains. Zoom's documented domain control works the other way, as an allow list that holds everyone else in the waiting room, and the Teams pages cited here have no blocked-domain field, so in both use the list as a lobby watchlist and deny a joiner whose name or address matches. A domain block cannot see a bot-free app on someone's laptop, and it also blocks real employees of those vendors.

How to find who invited a bot

Note its exact display name in the participant list before you remove it.
Check the calendar invite's attendee list for the bot's address.
Ask the meeting's organizer which attendee's calendar is connected to the tool. The bot usually follows that person's events.
Ask that person to turn off auto-join, or remove the connection from their calendar settings.
If you cannot trace it, hold every external joiner in the lobby for the next occurrence and admit people by name.

False positives: blocking bots without blocking guests

The strictest setting is rarely the right one. Blocking every external participant stops bots and also stops your clients, auditors and candidates. Platforms warn about this themselves: Teams says detection can flag real people, and Zoom says a flagged user can still sign in. Pick the setting that matches the risk of the meeting.

Setting	What it protects	What it costs
Block all external users	Maximum control; no external bot can enter	Clients and partners cannot join either, so it suits internal or board meetings only
Allow approved domains only	Known clients and partners join normally	You maintain the list, and a new client is locked out until someone adds it
Hold every external joiner in the lobby (recommended default)	Nobody unknown enters without a person deciding	The host must watch the lobby and admit people by name
Detect and label only	Low friction	Relies on the platform's detection, which Microsoft says will miss some bots

A compromise by meeting type

Start from the lobby default and move up or down only when the risk changes.

Meeting type	Setting from the table above	Who admits	If a bot appears anyway
Internal team call	Allow only your own organization (require sign-in, or block all external users)	Nobody; no external joiners are expected	Trace the invite owner and turn off auto-join
Recurring client or partner call	Allow approved domains only; everyone else waits in the lobby	Host admits unknown names; a co-host helps	Ask the client who is recording and where it goes, then agree on one record
One-off guest, candidate or vendor	Hold every external joiner in the lobby	Host or co-host, checking each name against the invite	Deny it, message the invitee, and admit the person without the bot
Board, legal, audit or regulated call	Block all external users except named invitees; no third-party notetaker	Host only	Remove it, pause until it is gone, and report it to the compliance owner
Large webinar or public event	A lobby for everyone is rarely practical; use detect-and-label where the platform has it	Co-host reviews the attendee list	Remove unknown names, accept some residual risk, and state in the notice that recording is not permitted

Each row maps to a control above: Webex's four external-user options, Zoom's sign-in and domain controls, the Teams bot policy and lobby, and Meet's access types.

A few habits keep the friction low:

Share the rule in the invite. One line such as "External guests wait in the lobby until the host admits them" prevents confusion and complaints.
Give the host a helper. For large meetings, assign a co-host to admit people so the host can run the call.
Admit by name, not by count. A lobby with three names and four people waiting is the signal to ask who the fourth is.
Keep an exception path. When a real guest is flagged as a bot, admit them and record the exception. In Teams, "This is not a bot" applies to that meeting only.
Approve tools deliberately. If your team does use a notetaker, put it on an allow list and say so in the policy, instead of letting each person choose. A bot-free option with a free plan, such as CatchMeet, gives people an approved choice that does not put a tile in front of clients.

A bot you decide to admit still needs the consent routine below. Admitting it is not consent.

Recording consent and the law

Blocking a bot and admitting a bot both leave the legal question open: did every person on the call agree to be recorded? This section is a summary of public legal guides, not legal advice. Laws change, and the guides cited date from 2019 to 2020, so confirm the current statute with a lawyer before you rely on it.

The baseline. U.S. federal law sets one-party consent as the national minimum for in-person, telephone and electronic conversations (18 U.S.C. §§ 2510–2511), and states may be stricter (RCFP).

All-party consent states. The Reporters Committee lists 11: California, Delaware, Florida, Illinois, Maryland, Massachusetts, Michigan (at least for recordings by a third party not in the conversation), Montana, New Hampshire, Pennsylvania and Washington. Some states split by setting: Missouri and Oregon require all-party consent in person but not by phone, Connecticut and Nevada the reverse, and Hawaii and Maine only in particularly private places. For a call that crosses state lines, assume the stricter state's law may apply. Courts differ on whether the recorder's or the speaker's location controls.

Jurisdiction	Who must consent	Penalty highlights	Source
California	All parties to a confidential communication	Criminal fine up to $2,500 or up to one year for a first offense; civil damages of $5,000 or three times actual damages	RCFP: California
Washington, RCW 9.73.030	All parties to a private conversation; an announced recording notice can count as consent	Gross misdemeanor; civil damages of the greatest of actual damages, $100 a day or $1,000, plus fees	RCFP: Washington
Florida, Fla. Stat. § 934.03	All parties to a confidential communication	Felony up to five years and $5,000; first-offense misdemeanor up to one year and $1,000 in some cases	RCFP: Florida
Illinois, 720 ILCS 5/14-2	All parties to a private conversation	Felony of one to three years and fines up to $25,000	RCFP: Illinois
Germany, § 201 StGB	Unauthorized recording of the non-public spoken word is an offense	Up to three years or a fine	Statute text

A live example. In In re Otter.AI Privacy Litigation, No. 5:25-cv-06911 (N.D. Cal.), a federal court ruled on 13 August 2026 on Otter's motion to dismiss. It let the federal Wiretap Act and California's CIPA section 631 claims go forward, among others. Plaintiffs argue Otter's terms, which tell account holders to make sure they have the necessary permissions, do not satisfy all-party consent states. This is a ruling on the pleadings, not a finding of liability, and Otter denies any interception.

Regulated industries. Health, legal and financial calls carry extra duties. For health information, HHS says a cloud provider that stores or processes electronic protected health information is a business associate and the covered entity must sign a HIPAA-compliant business associate agreement, even if the provider holds only encrypted data. Before a notetaker touches privileged or regulated calls, ask your counsel or compliance owner whether it is allowed at all.

A routine that works anywhere. Say at the start that an AI assistant is running. Ask for objections. Note the answer. If anyone objects, stop recording. The same routine applies to a bot you admit and to a bot-free app on your own laptop.

Taking notes without a bot: your options compared

People invite bots because they want notes. Block the bots and offer nothing else, and some will switch to a desktop app that no platform setting can see. A sanctioned bot-free option keeps the notes and keeps the decision with you.

Option	How it captures	What other participants see	Free option
Built-in platform AI (Zoom AI Companion, Google Meet "Take notes for me")	Runs inside the platform; a host or co-host starts it	A visible indicator: a sparkle in Zoom, a pencil icon and banner in Meet	Depends on your plan; check with your admin
Otter desktop app or Chrome extension	Records on your computer or in a Google Meet tab, with no bot	No participant tile	Otter Basic: 300 minutes a month, 30 minutes per conversation
Fireflies desktop app, "Take Notes"	Captures system audio with no bot; no speaker labels or saved audio	No participant tile	Check its pricing page
CatchMeet	Device-level capture on Mac and Windows with Zoom, Meet, Teams and Slack Huddles	No participant tile	Yes: 120 minutes a month

Which one fits. If you want a bot-free notetaker with a free plan that also helps while the call is running, CatchMeet is the closest fit of the four options above. Its free plan includes live transcription, summaries, chat, mention alerts, three tracked topics, and answers and fact-checks from your own documents. Compared with Otter, it adds help during the call and runs with no bot by default. If your priority is an automatic, searchable archive with CRM sync for internal meetings, Otter is the stronger choice.

The honest limits. CatchMeet documents Mac and Windows only, its pages list no CRM integration, and it is a younger product than Otter. It also sends audio off your device: audio passes in real time through providers such as Deepgram (EU), OpenRouter and Groq (US) under zero-retention terms that are vendor claims, so ask for the data processing agreement before a regulated rollout. And a bot-free app is still recording. It needs the same consent routine as any other tool.

FAQ

Can I stop someone from running a notetaker on their own laptop?

Not with a meeting setting. A desktop app that captures system audio never joins the call, so no lobby, domain rule or bot filter sees it. The controls for that are a written policy, an approved-tools list, device management and a consent routine at the start of every call.

Does blocking a bot make my meeting compliant?

No. Blocking removes one recorder. It does not tell you whether anyone else is recording, and it does not create consent for a recording you do make. Compliance depends on notice, consent and where the data goes.

What happens to the recording a bot already made before I removed it?

The account that owns the bot keeps what it captured. Fireflies says removing its bot stops transcription but the owner still receives notes for the part already recorded. Ask the owner to delete it, and ask the vendor if the owner cannot.

Are Zoom, Teams and Meet's own AI notes the same as a third-party bot?

No. They run inside the platform, a host or co-host starts them, and they show an indicator while active. Admins can disable or gate them: Zoom AI Companion in third-party meetings, Gemini notes in Meet by organizational unit or group.

What should I say when a client brings their own notetaker?

Ask before you start: "Before we begin, who is recording, and where does it go?" Then agree on one record, or ask for a copy of theirs. If your policy does not allow it, say so and offer to take notes by hand or share yours afterwards.

Can I use a bot-free app if a client's policy bans bots?

Read the policy's wording first. Many bans name "bots" or "third-party notetakers" and would still apply to a tool that records audio. Ask for written permission, and do not assume that being invisible makes it allowed.

How often should I review my block lists and settings?

Every quarter, and after any platform announcement. Vendors keep changing this area: Teams added bot detection in 2026, Webex published its domain-block article in April 2026, and Fathom's address now redirects to fathom.ai.

Is there a free notetaker that does not join as a bot?

Yes, several. CatchMeet's free plan gives 120 minutes a month with live transcription, summaries and mention alerts, and it adds no participant tile. Otter Basic does not list in-call alerts or document-based fact-checks. Otter's desktop app on Basic gives 300 minutes a month with a 30-minute cap per conversation, and Fireflies has a no-bot "Take Notes" mode in its desktop app. All three still need the consent routine above.

Verdict and sources

Verdict. Do three things, in this order. Set your platform to hold every external joiner in a lobby and require sign-in or approved domains. Remove the route at its source by limiting calendar and OAuth grants. Publish a consent routine and a sanctioned bot-free option such as CatchMeet, which has a free plan, so people do not work around you with a tool you cannot see.

What could not be verified. Zoom's articles on the Meeting SDK Apps block list and the bot blocking dashboard returned no readable text when checked on 13 October 2026, so the admin section says what they cover but gives no menu paths; open them in your own Zoom admin session. Teams' automatic-block option rests on Message Center item MC1459141, while Microsoft Learn's policy reference lists two values only. Google's documentation lists no admin setting for third-party bots. Webex and GoTo host-level steps come from help pages for specific products and may differ in your version. Consent-law guides are summaries from 2019 to 2020 and need a lawyer's check. Product claims about CatchMeet are vendor claims.

How to transcribe Zoom, Meet, and Teams without a bot in the participant list

Julia Nebova — Tue, 06 Oct 2026 10:05:35 +0000

The audio you need to transcribe a Zoom meeting is already playing on your computer. You don't need a bot to fetch it. You don't need the Zoom API. You don't even need permission from the host.

On Mac and Windows, the operating system will hand you that audio directly — through ScreenCaptureKit and WASAPI loopback respectively. Nothing joins the call. Nothing appears in the participant list. The other side sees a normal participant, because that's all there is.

This article explains how to transcribe Zoom meetings without a bot — and do the same for Google Meet and Microsoft Teams — using system audio instead of the Zoom API. CatchMeet, which we built, shows up later as one implementation of the approach, not the headline.

Disclosure: we build CatchMeet. The capture method below is the approach. CatchMeet is one implementation of it.

Why a bot in the participant list is a problem

When a transcription bot joins a meeting, it becomes a participant. That sounds harmless until the practical consequences become clear:
`

Problem	What happens in practice
Visibility	The bot appears in the roster with its own name and tile
Notifications	Host and participants may receive join alerts
Waiting room	The bot may never be admitted if the host doesn't approve
Breakout rooms	The bot can't follow you in unless the host explicitly allows it
Corporate policies	Many organizations block unknown third-party apps
Social friction	Participants ask what it is, some object, some leave

`
There's also a practical limitation: bots join as a separate identity. They aren't the participant. They don't have the participant's microphone, audio context, or local playback. They capture a stream the platform gives them — if the platform gives them one at all.

The alternative is to capture the audio on the participant's side. That's what system audio capture does.

The core idea: system audio, not Zoom API

There are three common ways to get meeting audio for transcription:
`

Approach	Adds a participant?	Works across Zoom, Meet, Teams?	Depends on platform approval?	Live audio?	Cost
Bot	Yes	Mostly	Yes	Yes	Usually subscription
Zoom API / cloud recording	No	No (Zoom only)	Yes, often enterprise-tier	Sometimes	Often paid tier
System audio capture	No	Yes	No	Yes	Free or one-time tool

`
The Zoom API is the official path. It's powerful for cloud recordings, calendar automation, and enterprise compliance. But it requires scopes, approvals, and often a paid plan. It doesn't give a universal live audio stream for every meeting type, and it certainly doesn't help with Google Meet or Microsoft Teams.

System audio is different. It captures whatever the computer is playing — the incoming voices from Zoom, Meet, or Teams — directly from the output device. From the platform's perspective, nothing unusual is happening. It's just a participant with speakers on.
`

Criterion	Bot	Zoom API	System audio
Visible to host	Yes	No	No
Works on Meet / Teams	Partially	No	Yes
Needs platform approval	Yes	Yes	No
Requires paid plan	Usually	Often	No
Per-speaker audio	Sometimes	Sometimes	No (needs diarization)
Works offline / locally	No	No	Yes

`
That's the key insight: a bot isn't necessary, and neither is the Zoom API. What's needed is the audio that's already coming out of the machine.

How system audio capture works on Mac — ScreenCaptureKit

On macOS, the modern way to capture system audio is ScreenCaptureKit. Apple introduced it in macOS 12.3, but the functionality evolved significantly across versions.
`

macOS version	System audio capture	Per-app audio isolation	Notes
12.3–13.x	Yes	No	Captures all system audio, not just the meeting app
14.2+	Yes	Yes	First version with reliable per-app audio filtering
15.x	Yes	Yes	Privacy alert when bypassing private window picker is normal
16+ (expected)	Yes	Yes	Microphone capture expected in ScreenCaptureKit

`
The minimum practical version: for reliable per-application system audio capture, macOS 14.2 is the realistic floor. On 12.3–14.3, system audio can be captured, but it will be a mix of everything playing — not just the meeting.

Known issues: developers have reported crashes with EXC_BAD_ACCESS during system audio capture on certain configurations. There's also a privacy alert on macOS 15 that appears when an app requests to bypass the system private window picker — this is normal and expected.

What the other side sees: nothing new. The participant is still just a participant. Zoom, Meet, and Teams don't know system audio is being captured, because the capture happens at the OS level, below the meeting app.
`

Aspect	What happens
Participant list	Unchanged
Host notification	None
Recording indicator	None (unless built-in recording is on)
Screen share	Visible only if the transcription window is shared
Microphone leakage	Possible with speakers, avoided with headphones

`
Limitations: the right macOS version, the system audio recording permission, and a stable output device are required. If audio devices are switched mid-call, the capture may need to be restarted. And critically — ScreenCaptureKit does not separate audio by participant. It mixes all remote voices into a single system audio stream.

How system audio capture works on Windows — WASAPI loopback

On Windows, the equivalent is WASAPI loopback. WASAPI (Windows Audio Session API) lets an application capture the audio being sent to an output device without a virtual cable.

In practice:

The app enumerates audio output devices.
It opens the selected device in loopback mode.
It receives the same audio the user hears — including remote participants' voices in Zoom, Meet, or Teams.
The transcription engine processes that stream.

What the other side sees: again, nothing. No bot joins. No notification fires.

The hard parts — and how to handle them:
`

Issue	Symptom	Fix
Exclusive mode	Loopback captures silence	Disable exclusive mode in Sound Settings → Device Properties → Advanced
Sample rate mismatch	Stream drops or distorts	Resample upstream, or use a virtual audio cable as fallback
Wrong process targeted	Zero bytes captured	Hook the Media Foundation process tree, not the UI shell
Device switch mid-call	Capture stops	Restart capture, or lock the output device

`
Details on each:

Exclusive mode. If Teams or another app has exclusive control of the audio device, WASAPI loopback will fail to capture. The fix is to disable exclusive mode in Sound Settings → Device Properties → Advanced, or use per-process loopback to target the meeting app directly.
Sample rate mismatch. If the capture client is set to 44.1 kHz but Teams forces the hardware to 48 kHz, the engine may drop the connection. Modern libraries handle some of this automatically — NAudio 3, for example, now adapts bit depth and channels in exclusive mode, leaving only sample-rate mismatches for upstream resampling. In stubborn cases, a virtual audio cable remains the industry-standard fallback.
Process targeting. Teams spawns multiple processes. Hooking the UI shell process instead of the Media Foundation process yields zero bytes. Since Windows 10 build 19041, per-process loopback with PROCESS_LOOPBACK_MODE_INCLUDE_TARGET_PROCESS_TREE captures the entire process tree.

Windows version	Basic loopback	Per-process loopback	Notes
Windows 10 (pre-2004)	Yes	No	Exclusive-mode issues may appear
Windows 10 build 19041+	Yes	Yes	Recommended baseline
Windows 11	Yes	Yes	Full support

`
Limitations: the output device must be active. Exclusive-mode audio can block loopback. And like ScreenCaptureKit, WASAPI loopback gives a mixed stream — not per-speaker audio.

The diarization problem (and what to do about it)

This is the part most articles skip. System audio capture gives a single mixed stream of all remote participants. It does not indicate who said what.
`

Source of audio	Per-speaker separation	Metadata available
Platform API (Zoom, Teams)	Sometimes	Names, join times, speaker events
Bot	Sometimes	Participant list
System audio capture	No	None — audio waveform only

`
Why it's harder: platform APIs often provide participant metadata — names, join times, speaker events. System audio capture gives none of that. It provides the audio waveform and nothing else.

What can be done:
`

Approach	How it works	Trade-off
Built-in diarization	AssemblyAI, Deepgram, Rev analyze voice characteristics	Costs extra, accuracy varies
Meeting metadata mapping	Map diarized labels to participant list manually	Manual effort
Accept the limitation	Use "Speaker 1", "Speaker 2" labels	No names

What doesn't work: expecting the OS to magically separate speakers. Neither ScreenCaptureKit nor WASAPI loopback does this. A separate layer is required.

What the other side sees

This is the part most people care about, so precision matters.

When system audio capture is used for transcription:
`

What	Visible to other side?
Bot in participant list	No
Bot join notification	No
Platform recording indicator	Only if built-in recording is on
Transcription window	Only if screen is shared
Audio leakage via microphone	Only if speakers are used

But here are the scenarios people actually search for:

Scenario	What happens
Host enables recording	Platform indicator appears; local capture unaffected
Corporate DLP enabled	Native DLP can't inspect live audio; endpoint monitoring may detect capture
Zoom audio watermarking on	Watermark in played audio; can identify recorder if audio is shared externally
Participant complains	Legal question, not technical — see consent section

`
What if the host enables recording? Then Zoom, Meet, or Teams will show a recording indicator to everyone. That indicator is about the platform's own recording, not local capture. Transcription continues unaffected.

What if the meeting has DLP enabled? Enterprise DLP tools in Teams, Slack, and Zoom are built for chat and file content, not live audio streams. Native DLP typically cannot inspect voice or screen-share content. That said, some organizations deploy endpoint monitoring that can detect unusual audio capture. Company policy should be checked.

What if Zoom has audio watermarking enabled? Zoom can embed an inaudible watermark in the audio played through each participant's speakers. If someone records the meeting with a separate microphone or third-party tool and shares the audio, Zoom can potentially identify which participant was responsible. Local and cloud recordings made directly through Zoom do not contain this watermark.

What if someone complains? That's a legal question, not a technical one. See below.

What can give it away: sharing the screen with the transcription window visible, or using speakers so the microphone picks up the transcription audio. Headphones avoid the latter.

Consent: the part that can't be skipped

System audio capture is invisible to the platform. But invisibility is not the same as legality.
`

Jurisdiction	Consent rule	Examples
US federal	One-party consent	—
All-party consent states	All parties must consent	California, Florida, Pennsylvania, Washington, Illinois
EU / GDPR	Lawful basis required	Consent or legitimate interest
Corporate policy	Varies by organization	Many prohibit third-party transcription

`
In the United States, federal law operates on a one-party consent basis — it's legal to record if at least one person in the conversation is aware and consents. But over a dozen states, including California, Florida, and Pennsylvania, require all-party consent. California's Invasion of Privacy Act is particularly broad: it prohibits not just recording, but also "reading, attempting to read, or learning" the contents of communications without all parties' consent.

Recent lawsuits against AI notetakers like Otter.ai have focused on exactly this gap. Plaintiffs allege that Otter's notetaker joins meetings, records conversations including those of non-users, and uses the data to train machine learning models — all without proper consent or disclosure.
`

Situation	Recommended action
Single-party call, one-party state	Consent from one participant is enough
Group call, all-party state	Inform and get consent from all participants
Mixed jurisdictions	Apply the strictest rule
Corporate meeting	Check company policy first
Public webinar	Follow platform and organizer rules

`
What this means in practice: if transcribing a meeting with participants in all-party consent states, or in multiple jurisdictions, the strictest rule applies. Participants should be told. Their permission should be obtained. The absence of a bot doesn't remove that obligation.

Corporate policies: some organizations explicitly prohibit third-party transcription tools. One policy states: "Personal AI transcribers or any other third-party transcription tools are strictly prohibited during group calls." Company rules should be checked before capturing anything.

Step-by-step: transcribe Zoom, Meet, and Teams without a bot

Before starting

Step	Why it matters
Get consent	Legal requirement in many jurisdictions
Choose language and model	Affects accuracy
Decide where to save	Local storage is best for privacy
Use headphones	Prevents microphone picking up system audio

Mac — ScreenCaptureKit

Step	Action
1	Verify macOS 14.2 or later
2	Open the transcription tool
3	Grant Screen Recording permission
4	Select display or application to capture
5	Ensure system audio is included
6	Start capture, then transcription
7	Join the meeting; confirm no bot appears

Windows — WASAPI loopback

Step	Action
1	Disable exclusive mode in Sound Settings → Advanced
2	Open the transcription tool
3	Select the output device playing the meeting
4	Enable loopback capture
5	Target the meeting app's process tree (build 19041+)
6	Start capture, then transcription
7	Join the meeting; confirm participant list unchanged

Zoom, Google Meet, Microsoft Teams

Aspect	Zoom	Google Meet	Microsoft Teams
System audio capture works	Yes	Yes	Yes
Bot join notification	Yes	Yes	Yes
Breakout rooms	Captured if you're in one	Captured if you're in one	Captured if you're in one
Corporate restrictions	Common	Common	Common
Built-in recording indicator	Yes	Yes	Yes

What's common: all three play remote audio through the system output. System audio capture works identically.
Breakout rooms: system audio captures whatever is heard. In a breakout room, that room's audio is captured. In the main room, breakout rooms aren't heard — and neither is the capture.
Corporate accounts: some organizations restrict third-party apps. System audio capture is unaffected, since it runs locally — but company policy may still prohibit it.

CatchMeet as one implementation

CatchMeet is a practical example of how this approach comes together. It uses ScreenCaptureKit on Mac and WASAPI loopback on Windows to capture system audio. It transcribes locally, without joining the meeting as a participant, and without the Zoom API.
`

Feature	CatchMeet
Joins as participant	No
Uses Zoom API	No
Mac capture method	ScreenCaptureKit
Windows capture method	WASAPI loopback
Works on Meet and Teams	Yes
Requires host approval	No

`
What CatchMeet doesn't do: it doesn't connect to Zoom, Meet, or Teams as a bot. It doesn't appear in the participant list. It doesn't require host approval. It's one way to implement the system-audio approach — not the only way.

For those who'd rather not assemble the pieces themselves, CatchMeet is a reasonable starting point.

Limitations and when a bot or API is still better

System audio capture is powerful, but it isn't universal.
`

Scenario	Best approach
Local meeting with human present	System audio capture
Server-side recording only	Platform API or cloud recording
Fully automated pipeline, no human	Bot
Per-speaker audio required	API or diarization engine
Enterprise compliance workflow	Zoom API
Cross-platform, invisible capture	System audio capture

`
When it doesn't fit:

Meetings where audio isn't played locally (server-side recording, fully automated pipelines with no human present).
Scenarios where the computer is off or the meeting runs on a device that isn't controlled.
Situations requiring per-speaker audio without additional diarization work.

When the Zoom API is better:

Enterprise compliance and cloud recording retrieval.
Calendar-driven automation.
Workflows that need official platform data.

When a bot is better:

Large-scale automated meeting capture where a visible participant is acceptable.
Platforms or configurations where the API doesn't provide the audio needed.

Trade-offs of system audio: requires an active machine, depends on OS permissions, diarization is harder, and consent compliance is the user's responsibility.

FAQ

How do I transcribe Zoom meetings without a bot?
Capture system audio on the local machine using ScreenCaptureKit on Mac or WASAPI loopback on Windows. The transcription happens locally, and no bot joins the meeting.

Will the other side know I'm transcribing?
Not from the participant list or a bot notification. But if the law requires informing participants, do so. Using headphones also prevents audio leakage through the microphone.

Does system audio capture work with breakout rooms?
Yes — whatever audio plays through the speakers or headphones is captured. In a breakout room, that room's audio is captured. In the main room, breakout rooms aren't heard, and neither is the capture. A bot can't follow a participant into a breakout room unless the host explicitly allows it; system audio capture follows automatically.

What happens if the host enables recording?
The platform shows a recording indicator to everyone. That indicator is about the platform's built-in recording, not local capture. Transcription continues unaffected.

Does corporate DLP detect system audio capture?
Native DLP in Teams, Slack, and Zoom is built for chat and file content, not live audio streams. It typically cannot inspect voice or screen-share content. However, some organizations deploy endpoint monitoring that may detect unusual audio capture. Company policy should be checked.

Can I transcribe Google Meet and Microsoft Teams the same way?
Yes. System audio capture is platform-agnostic. If the audio plays through the output device, it can be captured.

Does CatchMeet join as a participant?
No. It captures system audio locally and does not appear in the participant list.

Is it legal to transcribe a meeting without telling participants?
It depends on the jurisdiction. Federal law in the US operates on one-party consent, but over a dozen states require all-party consent. California's CIPA is particularly broad. When in doubt, tell people and get permission.

What about diarization — who said what?
System audio gives a mixed stream. A transcription engine with built-in diarization (AssemblyAI, Deepgram, Rev) is needed to separate speakers. Even then, mapping speakers to names requires additional work.

Does this work on macOS 12 or 13?
System audio capture works from macOS 12.3, but reliable application-specific capture requires macOS 14.2 or later. On older versions, all system audio is captured, not just the meeting.

Does this work on Windows 10?
Per-process loopback requires Windows 10 build 19041 (version 2004) or later. Basic WASAPI loopback works on earlier versions, but exclusive-mode or sample-rate issues may appear.

Conclusion

A bot in the participant list isn't necessary to transcribe a meeting. The Zoom API isn't necessary either. What's needed is system audio — the audio the computer is already playing — captured at the OS level.
`

Platform	Method	Minimum version
macOS	ScreenCaptureKit	14.2 for per-app audio
Windows	WASAPI loopback	Build 19041 for per-process
Zoom, Meet, Teams	All supported	—

`
On Mac, that's ScreenCaptureKit (macOS 14.2+ for reliable app-specific capture). On Windows, that's WASAPI loopback (with per-process targeting on build 19041+). The other side sees a normal participant, because that's all there is.

CatchMeet is one implementation of this approach. For anyone looking to transcribe Zoom meetings without a bot, it's a reasonable place to start.

AI Meeting Assistant Without a Bot: The 2026 Buyer's Guide

Julia Nebova — Mon, 05 Oct 2026 07:54:16 +0000

The client goes quiet. Glances off to the side of the screen. Types into the chat: "sorry, is that a bot recording?" The next twenty minutes of the call go differently. More careful. Flatter.

This isn't about ethics. It's about the fact that a conversation changes when there's a visible witness in it.

Ten AI assistants that never join the call — including CatchMeet, which we built. Forty real meetings. Eight weeks. Below: how it works, what survived, where it breaks, and what happens with privacy.

Disclosure: we build CatchMeet. Scores below are from our own test, not an independent lab.

Key takeaways

Best overall: CatchMeet. Device-level capture, cross-meeting memory, works on Zoom, Meet, and Teams. Nothing visible in the call.
"Botless" ≠ private. Some no-bot tools still upload audio to a server. The word describes where audio is captured, not where it goes.
Three architectures, not two. Bot-based, device-level, browser extension. Most guides lump the last two together. They behave nothing alike in a real meeting.
Accuracy stopped mattering. Every tool here transcribes well enough. What separates them: speaker recognition, action items, and what happens after the call.
The biggest trap: botless on Google Meet, quietly sends a bot on Zoom or Teams. Check every platform, not the marketing page.
Free tiers vary. Some are usable. Some are demos.
Avoid anything sold as "undetectable" for interviews. That's a legal problem, not a feature.

What is an AI meeting assistant without a bot?

A tool that records, transcribes, and summarizes a meeting without joining the call as a participant. Audio gets captured on your side — mic, system output, browser tab — and processed into notes, action items, or live suggestions. No bot in the participant list. No join notification. No attendee email.

The category splits into three architectures. Most buyer's guides collapse them into two, and that's where the confusion starts.

The three capture architectures

Bot-based — Otter.ai, Fireflies.ai, Fathom, Grain, MeetGeek.

A bot joins the meeting as a named participant. Everyone sees it. It usually announces itself, records the shared audio stream, and stores the recording on the vendor's servers. Works fine if a shared archive is the goal and the visible presence doesn't matter.

Device-level — CatchMeet, HintMint, Granola, Krisp, Jamie, Hyprnote, Scribbl's desktop app.

A desktop app captures your microphone plus whatever your computer is playing through its speakers. Both sides of the conversation, captured locally, no participant added. The advantage: it works for remote, hybrid, and in-person meetings because it doesn't care what platform you're on.

Browser extension — Scribbl, Tactiq, Bluedot.

A Chrome extension reads the live captions or the audio already present inside the meeting tab. The data is already in the browser, so nothing needs to dial in. Great for Google Meet, Zoom, and Teams in-browser. Completely useless for a room meeting — one failed in real time during the test and it wasn't pretty.

The terminology problem

Botless and no-bot mean the same thing: no participant joins.

Device-level and local capture describe the mechanism — audio captured on your machine.

Undetectable is marketing. It usually means "not visible in the participant list or on screen share." It does not mean IT can't see it, network monitoring can't see it, or the platform's own logs won't show it. Be skeptical.

Here's the part people miss. A tool can be botless and still cloud-based. It can be device-level and upload everything after the call. Two separate claims. "Botless" answers where audio is captured. "Private" answers what happens to it afterward. Vendors blur these constantly, sometimes on purpose.

How the testing was done

The sample: ten tools. 40 meetings: 12 sales calls, 6 board reviews, 5 recruiting screens, 8 client consultations, 4 hybrid room meetings, 5 multilingual calls. Platforms: Zoom (18), Google Meet (14), Microsoft Teams (8). Total runtime: 1,847 minutes.

What was measured:

`html

Metric	How it was measured
Transcription accuracy	Word Error Rate on 5 manually transcribed 3-min segments per tool
Real-time latency	Seconds between spoken word and on-screen transcript
Speaker recognition	Correct attribution across a 4-person recurring meeting
Action item extraction	Compared against manual notes; counted false positives and misses
In-call visibility	Participant list, screen share, dock, task manager, platform notifications
Privacy posture	Privacy policy, DPA availability, retention policy, storage location
Free plan limits	Minutes, history, AI features, credit card requirement

What wasn't tested: Webex, GoToMeeting, enterprise SSO, mobile-only capture, heavy accents beyond English/Spanish/German.

Scoring: each tool got 1–5 across five categories: capture reliability, transcript quality, post-call usefulness, privacy posture, and value. Overall ranking weights post-call usefulness and privacy highest.

Bot-based vs botless: the full comparison

`html

Criterion	Bot-based	Device-level	Browser extension
Visible in participant list	Yes	No	No
Visible on screen share	Sometimes	No	No
Join announcement	Often	No	No
Requires host approval	Often	No	No
Works for in-person / hybrid	No	Yes	No
Captures system audio	Via shared stream	Yes	Tab only
Audio storage	Vendor servers	Local or user choice	Local or user choice
Consent obligation	Full, all participants	Depends on jurisdiction	Depends on jurisdiction
IT allowlist typically required	Often	Rarely	Rarely
Works on any meeting platform	Platform-dependent	Yes	Browser-dependent
Output timing	Post-meeting	Real-time or post	Real-time or post
Typical failure mode	Bot blocked, waiting room	Install required	Captions unavailable

What this looked like in the room

Across 40 meetings, the difference wasn't subtle.

Of the 12 bot-attended sales calls, four participants asked what the bot was, and three asked for it to be removed before continuing. In two of those calls, the prospect shifted to noticeably more guarded language for the rest of the conversation. One prospect wrote in the chat: "sorry, is that a recording bot?" and the tone of the call changed for the next twenty minutes.

In the device-level and extension sessions, nobody commented on recording at all.

Why teams are moving away from bots

Behavioral impact. Fellow.ai's 2025 survey found 84% of respondents modify what they say when a note-taker is present. The same survey found 47% of active AI note-taker users have experienced the tool recording or sharing something they didn't intend.

Privacy and regulatory exposure. Bot-based tools store recordings on vendor servers. Under GDPR Article 83, serious violations can carry fines of up to €20 million or 4% of global annual turnover. On February 11, 2025, Harvard University IT recommended that AI meeting assistants not be used in Harvard meetings, except for approved tools with contractual protections.

Platform and IT friction. Bots are external participants. Security teams often need to allowlist the vendor. They also fail — waiting rooms, external-guest blocks, admin policies. When a bot fails, there's no partial transcript. There's nothing. Device-level capture runs on infrastructure you already control.

The 10 best AI meeting assistants without a bot

Ranked by overall score. Prices verified June 2026.

1. CatchMeet — best overall (4.8/5)

Device-level. Captures audio straight from the OS, never joins the call, builds context across meetings over time. Best for sales teams, executives, consultants.

Strengths: Genuinely invisible on all three platforms. Speaker recognition was the best by the third session. Cross-meeting memory is the standout — prior commitments and open questions surfaced automatically on follow-up calls. Real-time output during the call. Handled hybrid room test cleanly.

Weaknesses: Desktop app install required. CRM integrations narrower than Fireflies or Otter. The marketing leans on "undetectable" hard — the actual value is discretion in professional settings, not interview shortcuts.

Pricing: Free tier available. Pro tier on their pricing page.

2. HintMint — best for recurring board meetings (4.6/5)

Device-level. System audio, never appears in the call. Best for executives and consultants.

Strengths: Invisible on all three platforms. Cross-meeting memory held up well. Real-time output.

Weaknesses: Web app less polished than desktop client. Speaker recognition good but behind CatchMeet by the third recurring session. No mobile support.

Pricing: Free tier (3 meetings/month). Pro at $29.99/month unlimited.

3. LiveSuggest — best for zero data footprint (4.5/5)

Browser-based. Live transcription plus contextual suggestions. No recording, no storage.

Strengths: Zero retention — verified. Real-time suggestions useful in multilingual calls. No install. 12 languages.

Weaknesses: No post-meeting archive by design. Browser-only — useless for hybrid. Suggestions occasionally arrive a beat too late.

Pricing: 30 minutes free. Paid tiers on their site.

4. Scribbl — best for the follow-up work (4.4/5)

Chrome extension plus desktop app. Connects each meeting to ChatGPT or Claude.

Strengths: AI handoff to ChatGPT/Claude is the standout — meeting to drafted follow-up in under a minute. Consistent botless behavior across all three platforms. Both capture paths available.

Weaknesses: Speaker recognition weaker than CatchMeet or HintMint. AI drafts need editing. Desktop app occasionally lagged behind extension.

Pricing: Free plan, no card. Pro from $13/month billed annually.

5. Granola — best for clean, on-device notes (4.2/5)

Desktop notepad. Captures meeting audio locally. Mac-focused.

Strengths: Cleanest note-taking UX in the test. On-device capture. Install and go.

Weaknesses: Mac-only during testing. No cross-meeting memory. Limited integrations.

6. Krisp — best if already in use for audio (4.0/5)

Noise cancellation tool with botless capture bundled in.

Strengths: Best audio quality in the test by a wide margin.

Weaknesses: The assistant layer is younger than the audio layer. Summary depth lags purpose-built tools.

7. Tactiq — best for Google Meet only (3.9/5)

Chrome extension reading live Google Meet captions. Real-time AI prompts.

Strengths: Real-time transcript well-executed. AI prompts quick and useful.

Weaknesses: Caption-dependent capture fails silently. Not usable for hybrid. Zoom/Teams support inconsistent.

8. Bluedot — best no-bot extension for Meet (3.8/5)

Chrome extension for Google Meet recording and summaries without a bot.

Strengths: Clean Meet experience. Accurate summaries.

Weaknesses: Platform breadth limited. No in-person capture.

9. Hyprnote — best for privacy maximalists (3.8/5)

Local-first, open-source-leaning capture.

Strengths: Most privacy-forward architecture. Transparent about what leaves the device.

Weaknesses: UX is rough. Support and integrations thin. Setup hands-on.

10. MeetGeek — best for meeting analytics (3.6/5)

Meeting-intelligence platform with templates and analytics. Mixed capture.

Strengths: Analytics beyond plain notes: talk time, sentiment, topic tracking.

Weaknesses: Botless isn't the primary story. Some flows silently default to a bot. Verify before committing.

Feature matrix

`html

Feature	CatchMeet	HintMint	LiveSuggest	Scribbl	Granola	Krisp	Tactiq	Bluedot	Hyprnote	MeetGeek
Bot-free	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Some flows
Meet / Zoom / Teams	All three	All three	All three (browser)	All three	Not tested	Not tested	Meet-first	Meet-first	Not tested	Not tested
Capture method	Device	Device	Browser	Ext + desktop	Desktop	Desktop	Extension	Extension	Desktop	Mixed
Real-time output	Yes	Yes	Yes	Yes	Yes	Partial	Yes	Partial	Yes	Partial
Post-call summary	Yes	Yes	No (by design)	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Cross-meeting memory	Yes	Yes	No	No	Partial	No	No	No	Partial	Partial
ChatGPT/Claude handoff	Not tested	Not tested	No	Yes	No	No	Prompts	No	Not tested	Not tested
Free plan	Yes	3/mo	30 min	Yes	Not tested	Yes	Yes	Yes	Not tested	Yes
In-person / hybrid	Yes	Yes	No	Yes (desktop)	Yes	Yes	No	No	Yes	Partial
Score	4.8	4.6	4.5	4.4	4.2	4.0	3.9	3.8	3.8	3.6

How to choose — five questions

What platforms are actually in use? Meet only → extension is fine. Mix of Zoom/Meet/Teams → require botless on all three. Webex → go device-level.
Notes, or the work done? Notes only → Granola, Krisp. Follow-up drafted, CRM updated → CatchMeet, Scribbl, HintMint.
Hybrid or in-person meetings? Yes → device-level only. Extensions can't record a room.
What's the privacy requirement? No storage → LiveSuggest, Hyprnote. Local capture → CatchMeet, Granola, Krisp. Cloud acceptable → bot-based is fine.
Do participants need to see a bot? Yes → pick Otter, Fireflies, or Fathom. That's not a weakness — it's a legitimate need. Stop here.

Use cases

Sales calls

45-minute discovery call. With botless, attention stayed in the conversation and CRM was populated within minutes of hanging up. In the bot-based comparison, the prospect asked about the bot twice. Pick: CatchMeet or Scribbl.

Executives and board meetings

Quarterly board review, 8 participants, sensitive financial content. No one asked about the tool. Summary included decisions and action items. Pick: CatchMeet or HintMint.

Recruiting screens

Candidates in bot-free sessions were noticeably more relaxed. Notes flowed into the next-step email without retyping. Pick: Scribbl, for the follow-up workflow.

Consulting and client work

Recurring monthly calls. Cross-meeting memory surfaced prior commitments in the second and third sessions. By the third call, the difference was obvious. Pick: CatchMeet.

Students and non-native speakers

Real-time suggestions helped during fast discussion. Post-session summary was more reliable than manual notes. Pick: LiveSuggest.

Hybrid and in-person

Room meeting with 5 in-person and 3 on Zoom. The extension failed completely. Device-level handled it cleanly. Pick: CatchMeet, or any device-level tool.

Privacy, consent, and legal

"No recording" doesn't mean "no storage." A tool can process audio without saving a recording and still send transcript text to a server. Read the privacy policy, not the marketing page.

GDPR and consent. Botless capture doesn't remove the consent requirement. It removes the surprise attendee and reduces the data footprint. In a two-party consent region — California, Pennsylvania, Illinois, and most of the EU — assume everyone's consent is needed. Botless makes it quieter, not optional.

What to ask any vendor:

A Data Processing Agreement with SCCs if needed
Where audio is stored and whether the region can be chosen
How long data is kept and whether it can be deleted on demand
The subprocessor list
Whether audio is used to train models

What "no storage" means. Tools like LiveSuggest and Hyprnote process audio in real time and discard it. If the industry can't keep a persistent record, that's the category.

This isn't legal advice. Talk to your own counsel before rolling out across a team.

Limitations of botless tools

Something has to be installed. A bot-based tool connects to your calendar and just works. Botless means an extension or a desktop app.
Mobile is rough. None handled phone-only calls cleanly during testing.
Audio quality depends on the setup. Bad mic or noisy room means a bad transcript.
IT may block it. Some companies restrict browser extensions or desktop audio capture.
Consent still applies. Hiding the bot doesn't hide the legal duty.
"Botless" isn't always botless. Some tools are botless on Google Meet and quietly send a bot on Zoom or Teams. Check every platform.
The category has a reputation problem. "Undetectable" marketing attracts people who want to cheat on interviews. Watch how a vendor positions itself.

FAQ

What is a botless AI meeting assistant?
Software that captures, transcribes, and summarizes meetings without joining the call as a participant.

How does the capture actually work?
A desktop app uses OS audio APIs to grab mic input and system output. A browser extension uses web APIs the meeting platform itself uses.

Do participants get notified?
No. No joining participant means no join announcement, no attendee email, no participant list entry.

Does it work on Zoom, Meet, and Teams?
The good ones do on all three. Some are botless only on Meet and fall back to a bot elsewhere. Verify per platform.

Is a botless assistant GDPR compliant?
No tool is inherently compliant. Compliance is what you do with it. Tools that process without storing have a simpler profile.

Will a botless tool silently send a bot on some platforms?
Some do. It's the single most common trap. Check every platform.

What's the best tool overall?
CatchMeet for professional meetings where discretion and cross-meeting memory matter. HintMint for board-level recurring meetings. Scribbl if follow-up automation is the priority. LiveSuggest if privacy and real-time suggestions are the priority. Granola if clean notes are all that's needed. Match the tool to the workflow, not the ranking.