IT turned off calendar integrations for notetaker bots, or security flagged the one your team was using, and now nobody on that team has a working notetaker. The instinct is to look for a tool the block doesn't apply to. Before that, it's worth being precise about what the policy actually stops — because "no meeting bots" and "no unapproved recording software" are two different rules, and a tool built for the first one still needs to clear the second.
Current privacy behavior: New app versions offer optional usage statistics via PostHog, disabled by default. You can opt in during setup or in Settings and turn it off at any time. Audio, transcripts, notes, files, paths and search text are never sent. Older versions follow their embedded policy. See the current Privacy Policy.
What a company is usually blocking
A meeting bot is a separate account that joins a call as a participant — it shows up in the attendee list under its own name, and everything it records goes to that vendor's servers rather than staying with the company. Fireflies autojoins calendar meetings as fred@fireflies.ai by default once connected; Otter's AI Meeting Agent can be invited or set to autojoin the same way. Both patterns mean an outside company's software is dialing into an internal call and taking a copy of what's said. Admin consoles in Zoom, Teams and Google Meet can block third-party apps from joining meetings org-wide for exactly this reason, and IT teams turn that control on well before anyone on a team notices their notetaker stopped working.
Why "it doesn't join the call" answers one question, not every question
Coii AudioNotes doesn't request to join a meeting, because it isn't a participant in one — it records your own microphone and your Mac's system audio, the call as your own speakers are already playing it, entirely on the device you're using. There's no account for an admin console to block, because nothing dials in. That answers the specific mechanism a bot-blocking policy is usually built to stop.
It doesn't automatically answer a broader policy that says something like "no unapproved software may record meetings," full stop, regardless of whether that software joins as a participant or not. That kind of policy isn't really about bots — it's about which software is allowed to touch meeting audio at all, and a tool clearing the first concern doesn't get to skip the review process that answers the second one.
Why the block usually shows up without an announcement
Most people don't learn a policy changed from a memo — they learn it because a tool they were using quietly stopped working. A calendar-connected bot that used to join every meeting stops appearing one Tuesday, with no error message pointing at a policy, just an absent notetaker and a transcript that has to be written by hand instead. That's usually the admin console change, not a new company-wide announcement about recording software in general — security and IT teams often make org-wide changes to reduce risk without a broad rollout explaining each one, which is standard practice rather than a sign of anything unusual happening. The way to find out which policy actually changed is to ask directly, rather than guess from the symptom: "our meeting bot integration stopped working" and "we have a new policy against recording software" sound similar from the outside but call for different questions to IT.
Reading which policy you actually have
If the specific thing that stopped working was a calendar-connected bot no longer being able to autojoin — an admin turned off third-party meeting apps, or the vendor's integration got revoked — that's the narrower policy, aimed at the join mechanism itself. If the policy is written as "employees may not use unapproved software to record calls," that's the broader one, and it applies to any recording tool, local or cloud, bot or not, until it's specifically approved. The two read similarly from the outside — "my notetaker doesn't work anymore" — but they call for different next steps.
What to bring to whoever set the policy
Rather than assuming either policy does or doesn't cover a specific tool, bring the actual behavior to the person who owns the policy and let them tell you where it lands. The facts worth stating plainly, because they're the ones a security review actually checks:
- It records the Mac's own microphone and the audio the Mac is already playing — nothing from inside the call itself, and no separate participant.
- The recording, the transcript and the generated summary stay on that Mac. Nothing is uploaded anywhere.
- The only network activity is licence activation and a launch-time clock check — no analytics, no crash reporting, no calendar or CRM integration to authorize.
None of that is a claim that the tool is pre-approved, HIPAA-compliant, SOC 2-certified, or exempt from whatever process the company uses to approve software — it's the plain description a reviewer needs to make that call themselves, which is faster to give them than letting them assume it works like the bot they already blocked.
If the block is specifically about data leaving the company
A lot of bot policies exist because the recording leaves the organization's control the moment a third-party account joins the call — the transcript sits on a vendor's servers, subject to that vendor's retention and access rules rather than the company's own. That's a real, specific, checkable concern, and it's also the one a tool that keeps everything on the device answers most directly: there's no vendor server holding a copy, because there's no upload step in the first place. Say that plainly rather than arguing the bot policy shouldn't apply — the reviewer's actual concern is where the data goes, and "it doesn't go anywhere" is the direct answer to that concern, not a workaround for the policy that raised it.
If the block is about approved software generally
Some companies simply don't allow employees to install software on a managed Mac without it going through procurement or IT review first, independent of what the software does. In that case, the honest next step is the same one any other new tool on a work machine would need — get it added to the approved list, or ask whoever manages that Mac to install it — not looking for a category of software the policy happens not to mention yet. A tool that never joins a call as a participant is still a tool being installed on a company asset, and that's the part the general policy is actually about.
What this looks like for someone who needs notes today
If the immediate problem is a meeting later today and no approved notetaker, the honest options are limited to what's actually true: take notes by hand, ask a colleague to, or use whatever the company has already approved, even if it's slower. Installing something unapproved to solve today's meeting and sorting out the policy question afterward is a decision to make with eyes open about the order — reviews exist to happen before a tool touches company meetings, not after. If the bot itself is the immediate visible problem — one already showed up in a call and someone asked about it — that's a narrower, faster fix than the policy question: turning off autojoin for that account stops the next surprise while the broader review runs its course.
What "no bot" doesn't mean by itself
It's worth being clear about what not joining the call actually proves, so the claim doesn't get overstated in a review conversation. It proves there's no third-party account authenticating into the meeting and no vendor server receiving a live feed of the audio as it happens. It doesn't prove anything about how the recording is stored once it exists, how long it's kept, or who else on the same Mac could access it — those are real questions a thorough review will still ask, and they deserve real answers rather than being waved away because the bot question is already settled. Answering them honestly is part of the same conversation, not a separate one to avoid.
What a bot-free workflow actually looks like once it's approved
Once a tool clears whatever review applies, the day-to-day difference from a bot-based notetaker is mostly about what's absent: no bot joins the call to explain or dismiss, no attendee list entry to notice, no autojoin setting to manage per calendar. The other notetakers built the same way share that same absence, which is worth knowing if the review process cares less about a specific product than about the category of tool being requested.
Consultants and contractors working across several companies' policies
Anyone who works client-side rather than as an employee runs into this from the other direction: a consultant moving between two or three clients' calls in the same week may find one client's security policy is fine with a local recorder and another's is not, and the two answers don't generalize to each other. The behavior facts above are the same regardless of which client is asking, but the answer to "is this allowed" is specific to that client's own policy and their own reviewer, not something settled once for every engagement.
Regulated teams, where this question comes up hardest
Therapists, doctors and legal teams tend to have the strictest version of this policy, because the underlying data is the most sensitive kind a meeting can produce. What a regulated team should actually look for in a meeting notes app is the same behavior facts as above, described plainly rather than dressed up as a certification the app doesn't hold — the honest description of where the audio goes is what a compliance officer in that kind of role is equipped to evaluate, and it's the same conversation whether the blocked tool was a bot or something else entirely.
Once it's actually been reviewed
If a tool clears whatever review applies — IT adds it to the approved list, or a security reviewer signs off after seeing the behavior facts — that approval is worth keeping a record of somewhere the next person asking the same question can find it. Policies like this tend to get asked about repeatedly inside the same company, by different teams who hit the same blocked bot and go looking for the same answer independently, and a decision made once for one team is often treated as settled for the next one too, provided someone can point to it rather than starting the review from zero a second time. That's a smaller thing than the review itself, but it's the difference between one team solving this and the whole company solving it once.
What it costs
Coii AudioNotes is $19 once, for three of your own Macs, with a 30-day trial that needs no card and no account to start — useful for testing the actual behavior yourself before bringing it to whoever needs to sign off. It runs on macOS 13 Ventura or later, records nothing beyond what your Mac can already hear, and joins no call as a participant, which is the specific claim worth checking rather than assuming. See what a meeting bot actually is for the underlying distinction, or the case for meeting recorders that don't put a bot in the call for the wider argument this page is one piece of.