The short version: AI note-takers like Otter, Fireflies, Fathom and Zoom's built-in AI companion will save you hours a week on client calls, but they change how people talk to you the moment they know a bot is listening. Most small business owners switch one on without a script, without asking, and without a plan for where the transcript ends up. That's the bit worth fixing before your next call, not after.
What these tools do on a client call
An AI note-taker joins your Zoom, Teams or Google Meet call as a silent participant, records the audio, transcribes it, and then hands you a summary with action points, decisions and sometimes a highlight reel of quotes. Otter's free tier gives you 300 minutes a month, its Pro plan is around £14 a month. Fireflies runs roughly £10 to £19 per user per month. Fathom is free for the basic version and has become the one I hear recommended most by other consultants. Grain and Read.ai sit in the same space, aimed more at sales teams who want call analytics on top of the transcript.
The pitch is simple: stop scribbling notes, listen to your client, and get a tidy summary in your inbox five minutes after the call ends. For a one-person consultancy that's doing four or five discovery calls a week, that's two to three hours of admin gone. I'm not against the tools. I use one most weeks. But I stopped using them the way I used to.
The meeting that changed how I use these tools
Around eighteen months ago I was on a discovery call with a prospective client, a founder of a small e-commerce brand, and Fireflies was running in the background. We finished the call, said our goodbyes, and I hit the red button on Zoom to end the meeting for everyone. Except I hadn't left the room mentally, and neither had she. The call had ended for me but the bot had stayed on for another forty seconds because someone else on her side hadn't left the meeting window, and it kept transcribing.
What got captured was her turning to a colleague and saying something candid about her business partner that had nothing to do with our project and everything to do with why she was looking for outside help in the first place. It landed in my shared drive, tagged with her name, visible to a junior member of my team who had access to that folder for something completely unrelated. Nothing terrible happened. I deleted it, said nothing, moved on. But it sat with me for weeks, because it wasn't my information to have.
That's the moment I started treating the recording as something that needed managing, not something that just happened in the background.
The trust problem other guides skip over
Here's the bit most articles about AI meeting tools don't say out loud: people give you worse, safer, more rehearsed answers once they know a bot is capturing every word. The honest objection, the "look, between you and me, the real problem is my co-founder" moment, the actual budget number instead of the one they lead with, these tend to come out after someone says "right, shall I stop the recording there?" If your AI note-taker is running the whole time, you often get a perfectly clean transcript of a conversation that never got to the real substance.
I've watched this happen from the other side too, sitting through supplier pitches where I could feel the person on screen choosing their words like they were being deposed. Nobody says this because it makes the tools sound less useful than the marketing suggests. But a transcript of a guarded conversation isn't more useful than a rough handwritten note of an honest one. It's just tidier.
There's a version of this that brands get right on a bigger scale. Look at how Airbnb built its brand on trust after years of horror stories about guests and hosts, or how Away built its whole customer relationship on being upfront about what you're getting before you buy. Both companies learned that being transparent about the mechanics of a transaction, not just the benefits, is what makes people relax into it. A recording bot on a client call is a small version of the same problem. Say what it's for, say where it goes, and people talk to you like a person again instead of a witness.
The rules you need to know in the UK
Recording a call counts as processing personal data under UK GDPR the moment there's a person's voice, name or identifiable detail in it. You need a lawful basis to do that, and "I forgot I had it switched on" is not one of them. In practice, for most small business consultations, consent is the easiest lawful basis to point to, which means telling the other person before the call starts, not slipping it into your calendar invite in six-point font.
A few practical points that came out of me reading the guidance from the Information Commissioner's Office rather than trusting a LinkedIn post about it:
- Tell people verbally at the start of the call, not just in writing beforehand. "I've got Fireflies recording this for my own notes, that alright with you?" takes four seconds.
- Don't store transcripts indefinitely by default. Most tools keep everything forever unless you set a deletion policy, which almost nobody does.
- If a call touches on anything sensitive, health, finances, a personal dispute between business partners, that's a moment to switch the recording off, not a moment to let the AI capture it and hope for the best.
- Anyone on your team with access to the shared transcript folder needs an actual reason to be in there. Default sharing settings on most of these tools are far too open.
How I introduce it now
I changed my script after that call and it's stayed the same since. At the top of every discovery call I say something close to: "I use an AI note-taker so I'm not scribbling while you're talking, it'll send me a summary afterwards, nothing goes anywhere else, are you happy with that?" That's it. Nobody has ever said no. But two or three people a month ask me to switch it off partway through, usually right before they tell me something that matters, and I do it without hesitation because that's the whole point of asking in the first place.
I also stopped letting the bot join calls with existing clients where we're discussing anything contractual or where there's tension in the relationship. Those calls I take notes myself, badly, by hand, the old way. It's slower. It's also more honest, because people know I'm the only one hearing them.
Where the real time saving shows up
The saving isn't in the meeting itself, it's in what happens after. A one-hour discovery call used to cost me another twenty minutes writing up notes and action points. With a transcript tool doing the first draft, that's down to about five minutes of editing. Across a week of five client calls, that's roughly an hour and a quarter back, every single week, which over a year is somewhere near sixty hours. That's the actual case for these tools, not "never miss a word," but "stop losing your Friday afternoon to admin."
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.
Where it goes wrong is when business owners treat the transcript as the deliverable rather than the starting point. A forty-minute call generates about six pages of transcript. Nobody, including you, is reading that back in three months. What you need is the three-line summary and the two action points, and every tool I've mentioned will give you that automatically if you set it up right the first time rather than accepting the default template.
Where this fits with the rest of your client experience
None of this exists in isolation. If you're already using an interactive quiz or an interactive calculator to qualify leads before they ever book a call, an AI note-taker is just the next tool in the same chain, one more piece of the client's data trail that they need to know about. The businesses that handle this well tend to be the ones who are blunt about everything else too. Alex Hormozi built a chunk of his reputation on saying the uncomfortable thing out loud before anyone else would, and there are a few useful business lessons from Alex Hormozi in there about not dressing up a transaction as something friendlier than it is. Ahrefs did something similar in content marketing, building an entire audience by being straightforwardly useful rather than polished, which is worth a look if you want to see how Ahrefs built a brand people trust without ever softening their message.
If you're running a small team and want someone to sit down and map out which parts of your client process should have AI in them and which shouldn't, that's a fairly specific conversation, and it's exactly the kind of thing an AI implementation coach works through with a business in the first week, not the tenth.
My working rule now
Record the calls where a clean summary helps both sides. Turn it off the second a conversation gets personal, contractual, or tense. Tell people what's running before you start, every time, without exception. And don't mistake a tidy transcript for a good conversation, because they're not the same thing, and the businesses that forget that end up with beautifully organised notes of meetings where nobody said anything true.
Frequently asked questions
Do I legally have to tell clients I'm recording a call with an AI note-taker?
In the UK, yes, in practical terms. Recording a call involves processing personal data under GDPR, and consent given verbally at the start of the call is the simplest lawful basis to rely on. Telling people in a calendar invite footer they never read doesn't count as proper consent.
Which AI note-taker is best for a small business on a budget?
Fathom's free tier covers most solo consultants and small teams without any cost. Otter's free plan gives 300 minutes a month, which suits anyone doing three or four calls a week. Fireflies and Grain add more analytics but start costing real money once you add team seats, usually £10 to £19 per user monthly.
Does using an AI note-taker change how clients behave on calls?
Yes, and it's rarely mentioned in the marketing for these tools. People tend to give more guarded, rehearsed answers once they know a bot is transcribing everything, and the most honest part of a conversation often happens right after someone asks you to stop recording.
Should I let an AI note-taker join every single client meeting?
No. It's useful for standard discovery calls and project updates where a clean summary saves you admin time. Switch it off for anything contractual, anything personal, or any call where there's tension in the relationship, because those are the moments where people need to know only you are listening.
For the bigger picture, see my full guide to AI marketing.
Related reading: AI Meeting Note-Takers: What Nobody Tells You Before You Turn One On and The AI Notetaker Problem Nobody's Warning Small Business Owners About.
