The short version: AI notetakers like Otter, Fireflies and Fathom do save time on client calls, but not as much as the marketing suggests once you count the editing, and most small business owners are using them without telling clients they're recording, which is both a trust problem and, in the UK and EU, a legal one. Fix it with a fifteen second consent line before every call and a written policy, not a blanket ban.
The call where a client stopped everything to ask who was listening
I was on a discovery call with a coaching client a few months back, let's call her Ruth, and about four minutes in she stopped mid sentence and said "sorry, who's Otter?" There was a little bot sitting in the Zoom participant list, quietly transcribing. I'd forgotten I'd switched it on. She wasn't angry, but she was clearly uncomfortable, and she asked me to turn it off before she'd carry on talking about her business finances.
That five minute wobble taught me more about AI notetakers than a year of using them had. The tool itself wasn't the problem. The silence about it was.
What these tools do (and how fast they've spread)
If you're not using one yet, you're in a shrinking minority. Otter.ai, Fireflies.ai, Fathom, Read.ai and Grain all do roughly the same job: a bot joins your Zoom, Teams or Google Meet call, transcribes it in real time, then hands you a summary, action items and sometimes a highlight reel. Fathom's basic plan is free. Fireflies runs around $10 a month per user. Otter's Pro plan is about $16.99 a month. None of that is expensive, which is exactly why adoption has been so quick.
I run twelve to fifteen client calls a week between consulting, coaching and podcast guesting. I use a notetaker on most of them now. But I've stopped pretending it's a pure time win, because it isn't.
The maths nobody publishes
Here's what I track, not what the sales page for the tool tells you:
- Writing my own notes during a 45 minute call used to take about 10 minutes afterwards, because I type fast and I know what matters.
- A notetaker hands me a transcript and an AI summary in about 90 seconds after the call ends.
- But editing that summary, because it misattributes quotes, flattens nuance, or invents an action item that was a joke, takes me 12 to 15 minutes on a typical call.
So the "time saved" is often two to three minutes, not the "hours back in your week" line every one of these tools uses in its ads. Where it does pay off is on calls with three or more people, where I could never have caught everything myself. That's the honest use case. Solo one to one calls, less so.
I wrote about a similar gap between the promise and the reality when I published 569 blog posts in 45 days using AI and Google only showed 5 percent of them. Same pattern: the output looks finished, but the finishing is where the real time goes.
The consent problem most small business owners are ignoring
This is the part that makes people uncomfortable, so most posts about AI notetakers skip it. In England, Wales and Northern Ireland, you can legally record a call if you're a participant in it, even without telling the other person, under the Regulation of Investigatory Powers Act. But recording is not the same as processing. The moment that recording becomes a transcript stored on a US server, analysed by an AI model, and potentially used to train that vendor's systems, you're into GDPR territory, and GDPR requires a lawful basis and, in most business contexts, clear notice to the other person. The GDPR doesn't care that the "notetaker" felt harmless.
Scotland's rules on covert recording are stricter still. And in several US states, including California, Florida and Illinois, you need consent from every person on the call, not just yourself, or you're breaking the law outright.
None of this means don't use the tools. It means tell people. I now say one sentence at the start of every call: "Quick heads up, I've got an AI notetaker running so I don't miss anything, are you happy with that?" In two years, exactly one person has said no. Ruth,. And that's fine. I turned it off and took notes the old fashioned way for that whole relationship.
Why clients notice more than you'd think
Small business owners assume clients either don't notice the bot or don't mind it. Neither is reliably true. If you've spent years building trust with clients as a non-technical business owner using AI day to day, the thing that damages that trust isn't the AI itself, it's finding out later that something was recorded and you weren't told. I've had two people mention, weeks after a call, that they'd googled the notetaker's name because they'd seen it in the participant list and wanted to know where their words were going. That's not paranoia. That's a completely reasonable question, and if you can't answer it clearly, you've got a problem bigger than a missed action item.
Here's the uncomfortable bit most AI content on this topic avoids: the tools are being sold to you as a productivity win, but for a lot of small businesses the real function they're serving is a paper trail. If a client disputes what was agreed, you now have a transcript. That can be useful. It can also be used against you just as easily, and it changes the tone of a call the moment someone knows every word is being kept.
A simple policy you can put in place this week
You don't need a lawyer for this, you need six lines in your onboarding document and a habit. Here's what I run:
- State it verbally at the start of every call, every time, no exceptions, even with regular clients.
- Put a line in your client contract or welcome email: "Calls may be recorded and transcribed using AI notetaking software for accuracy and follow up purposes."
- Offer an opt out and mean it. If someone says no, turn it off, don't argue.
- Delete transcripts you don't need after a set period. I clear mine every 90 days unless there's an active project reason to keep one.
- Never let the AI summary go out to a client unedited. Read it, fix the parts it got wrong, then send it.
- Check where the vendor stores data and whether they train their model on your calls. Fathom and Fireflies both let you opt out of model training in settings; check that box.
That's the whole policy. It took me longer to write this paragraph than it did to implement.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
When it's worth using one
Group calls with three or more stakeholders, where things get agreed quickly and nobody's assigned to take minutes. Sales calls, so you can review your own pitch later and hear where you rambled. Podcast interviews, where a rough transcript saves an editor hours. Discovery calls with new clients, where you're absorbing a lot of detail fast and you will forget half of it by Thursday.
Where I'd skip it: one to one coaching calls involving anything personal or financial, calls with clients who've already shown discomfort with tech, and any call where the other person is clearly more senior or cautious than you and hasn't been asked in advance. In those cases, a notepad and a pen still work fine, and they always will.
The bit that has nothing to do with the AI
The reason this whole thing matters more than it looks is the same reason I've been rebuilding my own business in public for the past five years. Small stuff like this, a bot joining a call uninvited, a summary sent with an error in it, a client finding out something after the fact, is what quietly erodes trust long before anything big goes wrong. I've written before about the month my website started paying me again, and the pattern was the same there too: it wasn't one big fix, it was a string of small, unglamorous, slightly boring decisions done consistently. Telling a client you're recording the call is one of those decisions. It costs you nothing and it's the difference between a tool that helps you and one that quietly works against you.
If you want to see how far the AI tooling ecosystem has spread beyond the obvious names, have a look at this list of lesser known but useful websites, several of the newer meeting and workflow tools on there are worth a look precisely because they're not household names yet, which usually means better support and fewer data privacy headaches while they're still building trust with early users.
Free resource: The Discovery Call Note-Taking Template.
Frequently asked questions
Is it legal to use an AI notetaker on a client call without telling them?
In parts of the UK it's legally allowed to record a call you're a participant in without announcing it, but processing that recording with AI and storing the transcript almost always brings GDPR into play, which requires a lawful basis and usually clear notice. In several US states, including California and Illinois, you need consent from everyone on the call or you're breaking the law. Tell people. It's one sentence and it removes the risk entirely.
Do AI notetakers save small business owners time?
Sometimes, not always. On group calls with three or more people they're a genuine time saver because no human could catch everything. On solo one to one calls, the editing time to fix errors in the AI summary often eats most of the time you saved, so the net gain can be as small as two or three minutes.
Which AI notetaker is cheapest for a small business?
Fathom has a useful free plan. Fireflies runs around $10 a month per user. Otter's Pro plan is about $16.99 a month. For a solo consultant doing under twenty calls a week, the free or lowest paid tier of any of these is usually enough.
Should I ever say no to a client who wants to use an AI notetaker on our call?
Yes, if you're not comfortable with it. You're entitled to ask for it to be switched off, especially on calls involving anything financial, medical, or personal. Any business worth working with will turn it off without an argument.
Want this done for you? See AI automation for coaches and consultants.
Related reading: AI Notetakers Are Quietly Costing You Client Trust. Here's What I Do Instead and Why I Turned Off My AI Meeting Notetaker After 14 Months (And What It Was Really Costing Me).
