
Your notetaker used to get into every call. Now a customer's Microsoft Teams meeting can flag it as a suspected bot and hold it in the lobby until someone lets it in.
If you run recording for a sales or CS team, some calls still need a bot and others need a recorder that never joins. This guide compares both on speaker data, consent and reliability, then gives you a meeting-by-meeting plan to turn into a team policy.
TL;DR
Bot-based recording means a notetaker joins your Zoom, Microsoft Teams or Google Meet call as a separate participant. It gets invited from your calendar, shows up in the participant list, and records what the platform sends it.
Botless (or bot-free) recording means the capture happens on your device. A desktop or mobile app records your microphone and the audio your computer plays, so no one new appears in the call.
There are 4 common botless routes:

Where the capture happens shapes everything else, from who sees the recorder to how speakers get labeled.
The two methods differ most on speaker labels, attendance and disclosure.
| Dimension | Bot-based recording | Botless recording |
|---|---|---|
| How it records | Joins the call as a participant | Captures audio on your device |
| What participants see | A named notetaker in the participant list | No extra participant |
| Setup | Joins from your calendar invite | You start it, or the app prompts you |
| Meetings you don't attend | Records without you | Needs you and your device in the meeting |
| Speaker labels | Can use each participant's separate audio | Your mic plus one mixed track for everyone else |
| Where it works | Zoom, Teams, Meet and other supported platforms | Any audio your device plays, plus in-person on mobile |
| Typical failure | Blocked by the host's IT or stuck in the lobby | Nobody presses record, wrong mic, laptop asleep |
| Where audio goes | The vendor's cloud | The vendor's cloud, unless the tool is local-first |
| Consent | Required | Required |
The blocked-bot row is getting more common. Microsoft Teams now uses bot detection to hold outside notetakers in the lobby until an organizer admits them, and Microsoft is rolling out an option to block detected bots.
For revenue teams, the speaker-labels row matters most, and the next section shows why.
Speaker attribution is the main difference. Word accuracy depends on audio quality, but knowing who said each sentence depends on how the audio arrives.
A bot inside Zoom or Teams can often receive each participant's audio as a separate stream. Recall.ai, which builds recording infrastructure for meeting tools, explains in its diarization documentation that separate streams make speaker attribution more accurate than one mixed track.
A desktop recorder hears your mic and your speakers. Your voice arrives on its own, but everyone else arrives mixed, and software has to separate them by sound.
Revenue workflows depend on who said what:
Example: On a botless discovery call with 3 buyers, the transcript separates the rep from the buyers. Telling the CFO from the champion is harder, and that decides who raised the pricing objection. (Illustrative scenario.)
Pro Tip: Before moving a team to botless, record one internal multi-person call both ways and compare the speaker labels.
Speaker data is the main cost of going botless. Consent is the main risk.
Consent obligations stay the same with botless recording. Recording laws apply to the act of recording, whichever way the audio gets captured.
In the US, some states allow one-party consent, and others require all-party consent. A call spanning several states or countries can fall under more than one set of call recording laws.
A visible bot is a cue, but it doesn't prove anyone agreed, and botless recording removes even that cue. Lawsuits against AI notetakers, including a class action against Otter.ai, have made consent a standard question in any AI notetaker security review.
How each method handles disclosure:
Without an announcement feature, say it in the first 30 seconds:
"Quick note before we start: I'm recording this call so I can focus on our conversation. Is that okay with everyone?"
If anyone says no, stop recording and take notes by hand. Treat this as general information, and check the rules for each location with your legal team.
Common Mistake: Treating an accepted calendar invite as consent. Accepting means agreeing to attend, and recording needs its own notice.
With speaker data and consent covered, you can pick a method per meeting.
Default to a bot for scheduled calls where you need coaching data or might not attend. Use botless when a bot can't get in or isn't welcome.
The 3-question recording check runs in this order:
Whichever method you land on, tell people they're being recorded.
| Meeting type | Recommended method | Why |
|---|---|---|
| Discovery and demo calls you host | Bot, with botless as the fallback | Speaker data for coaching and CRM; switch if the buyer objects |
| Calls hosted by a customer that blocks bots | Botless | The bot won't get past their lobby |
| Sensitive calls (exec, legal, pricing) | Botless with spoken consent, or no recording | No extra participant; ask first |
| Internal 1:1s and team meetings | Bot | Records without you and labels speakers |
| Meetings you can't attend | Bot | A device recorder needs you in the meeting |
| In-person meetings | Botless (mobile or desktop app) | No meeting link for a bot to join |
| Phone calls | Dialer integration, or botless with an announcement | Meeting bots can't join phone calls |
| Slack huddles and WhatsApp calls | Botless | Meeting bots can't join these |
| Impromptu calls with no invite | Botless | No calendar invite for a bot |
A matrix only helps if reps know when to switch, which is what a written policy is for.
Avoma records both ways. The Avoma Assistant joins scheduled Zoom, Microsoft Teams and Google Meet calls, and Teams admins can set how lobby approval treats it. The Avoma desktop app records any call without a bot, from the menu bar on Mac or the system tray on Windows.
The desktop app runs on Windows and Mac (Apple silicon) and also shows live notes and coaching during calls.
The calls your bot can't reach still hold deal context worth keeping. Download the Avoma desktop app and start with the botless rows from the matrix above.
Use Avoma's Desktop App to record calls when meeting bots can't join or aren't welcome. Get AI notes, transcripts, and conversation insights without adding a bot to the meeting.


Use Avoma's Desktop App to record calls when meeting bots can't join or aren't welcome. Get AI notes, transcripts, and conversation insights without adding a bot to the meeting.

Use Avoma's Desktop App to record calls when meeting bots can't join or aren't welcome. Get AI notes, transcripts, and conversation insights without adding a bot to the meeting.

Use Avoma's Desktop App to record calls when meeting bots can't join or aren't welcome. Get AI notes, transcripts, and conversation insights without adding a bot to the meeting.

Use Avoma's Desktop App to record calls when meeting bots can't join or aren't welcome. Get AI notes, transcripts, and conversation insights without adding a bot to the meeting.

Use Avoma's Desktop App to record calls when meeting bots can't join or aren't welcome. Get AI notes, transcripts, and conversation insights without adding a bot to the meeting.


Use Avoma's Desktop App to record calls when meeting bots can't join or aren't welcome. Get AI notes, transcripts, and conversation insights without adding a bot to the meeting.

Use Avoma's Desktop App to record calls when meeting bots can't join or aren't welcome. Get AI notes, transcripts, and conversation insights without adding a bot to the meeting.


Yes, with a desktop or mobile app. Device-based recording captures the audio a computer plays, so it works whether or not the user hosts the meeting. Platform recording, such as Zoom cloud recording, depends on the host's permissions and plan. Consent rules apply either way.
Only if someone tells them. A botless recorder adds no attendee to the participant list. Some desktop apps play a spoken recording notice through the computer's audio. Without that feature, the person recording needs to tell participants at the start.


