Straight answer: AI notetaker bots (Otter, Fireflies, Fathom, Read) can save you hours a week, but most small business owners join calls with them switched on by default and never tell the other person what's recording them, which in plenty of places is not just bad manners, it's a breach of consent law. Fix the disclosure, and the tool is brilliant. Skip it, and you're one annoyed prospect away from a complaint.
Useful alongside this: How An AI Meeting Assistant Saves You Time On Note-Taking (And Where T.
Worth reading next: The AI Note-Taker Nobody Invited: What It's Doing With Your Client Cal.
What these bots do (and why everyone's suddenly using one)
Eighteen months ago almost none of my calls had a third guest sitting silently in the corner of the screen. Now most do. Otter.ai, Fireflies, Fathom, Read, Grain, tl;dv, they all do roughly the same job: join your Zoom or Teams or Google Meet call, transcribe every word, summarise the action points, and file it away searchable forever.
The pricing tells you how fast this has moved from novelty to normal. Fathom is free with unlimited meetings, which is why it's eaten a huge chunk of the market in under two years. Fireflies runs about $10 to $19 a month per seat depending on the tier. Otter's Business plan is roughly $20 a seat monthly. For a one-person or five-person business, that's nothing against the hours you'd otherwise spend typing up notes after every call.
And the notes are useful. I've gone back to a transcript three weeks after a discovery call to check the exact number a prospect gave me for their current ad spend, because I misremembered it in my own head and would have quoted the wrong figure back to them. That's a real, concrete win. The problem isn't the technology. It's how casually small business owners are deploying it.
The call that changed how I think about this
I was on a discovery call last year with a prospect, a director at a mid-size manufacturing firm, and about four minutes in a bot named something like "Fireflies Notetaker" popped into the participant list. I hadn't invited it. The prospect hadn't mentioned it. When I asked, mildly, whether he minded that I make my own notes, he laughed and said "oh don't worry, mine's already recording, it does that automatically now, I don't even think about it."
He didn't think about it. That's the bit that stuck with me. He had no idea the transcript was about to land in his shared workspace, visible to three colleagues, one of whom (he told me later, unprompted) was actively talking to a competitor of mine about the same brief. He hadn't chosen to record me. His calendar app had made that choice for him, on autopilot, the same week he'd upgraded his Google Workspace plan.
Nobody asked me. Nobody asked him, either, not really. That's the uncomfortable bit most people writing about "AI productivity hacks" skip over: consent has quietly become opt-out instead of opt-in, and almost nobody running a small business has checked whether that's legal where they operate.
The consent problem, in plain terms
In the UK, recording a call without telling the other party isn't automatically illegal, but under UK GDPR you're processing someone's personal data the moment you transcribe their voice, and you need a lawful basis and a way for them to know it's happening. The Information Commissioner's Office is clear that covert recording of business calls is something you need to actively manage, not something you can wave away because "the software did it." If you want the source rather than my paraphrase, the ICO's own guidance spells out what counts as personal data processing on a call.
In parts of the US it's a much bigger problem. States like California, Florida, and Pennsylvania are "two-party consent" states, meaning both people on the call have to agree to being recorded, not just one. If you're a small business owner in London selling to a client in Los Angeles and your notetaker bot joins uninvited, you could be in breach of California law without ever having read a word of it.
I'm not a lawyer and this isn't legal advice. But here's the plain business version: if your calendar app auto-invites a notetaker to every call and you've never once said out loud "I'm recording this for notes, is that alright," you have a gap. Close it. It takes one sentence.
The bit nobody puts in the "top 10 AI tools" listicles
Here's the part that's harder to admit: AI notetakers change how people talk. I've noticed it in myself. Once I know a transcript is being generated, I speak in tidier sentences. I reach for the soundbite instead of the honest, messy answer. Prospects do the same thing back at me. The call gets slightly more careful, slightly more rehearsed, and a chunk of the rapport that used to build in the scrappy, off-the-cuff moments just doesn't show up anymore.
That matters more in sales and consulting than almost anywhere else, because trust is built in the unguarded seconds, not the polished ones. A discovery call where both people are performing for a transcript is a worse discovery call, even if the notes afterward look immaculate. The tool is optimising for record-keeping and quietly taxing the human connection that got you the meeting in the first place.
Companies that understand this tension tend to be explicit about it rather than hiding the mechanism. Look at how Hotjar built its entire marketing strategy around being upfront that it records website visitor sessions, telling customers exactly what's captured and why, instead of letting it run silently in the background. Small businesses using AI notetakers could learn from that: say the recording word out loud, and most people relax about it in seconds.
How to use an AI notetaker without the risk
This is the bit that took me about twenty minutes to fix once I'd noticed the problem, and I'd rather you didn't spend eighteen months not noticing it like I did.
- Turn off auto-join. In Otter, Fireflies, and Fathom you can switch the bot from "joins every meeting automatically" to "joins only when I add it." It's one toggle in settings. Do it today.
- Say the sentence. "I use an AI tool to take notes so I don't miss anything, is that alright with you?" Takes four seconds. Nobody has ever said no to me.
- Rename the bot. Default names like "Fireflies Notetaker" look like spam bots in a participant list. Rename it something plain like "Lilach's Notes" so it reads as a tool you're using, not a stranger who wandered in.
- Set a deletion rule. Most tools let you auto-delete transcripts after 90 or 180 days. Do this, especially for calls that touch on client financials or anything sensitive.
- Check where the transcript lands. If your plan shares transcripts with a whole team workspace by default, lock that down to the people who need it, particularly if you're talking to more than one client in the same niche.
Five steps, none of them technical, all of them the difference between "useful tool" and "quiet liability."
Where AI notetakers earn their keep
Once the consent bit is sorted, the actual value is real and worth being specific about. On a typical week of six client calls, I used to spend roughly forty-five minutes after each one writing up notes and action points, call it four and a half hours a week. With a transcript and an AI-generated summary doing the first pass, that's down to about fifteen minutes of me checking and correcting the summary, because the tools still get names, numbers, and context wrong more often than any vendor admits.
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.
That's roughly three hours a week back, every week, for a solo consultant. Multiply that across a small team and you're talking about a genuine chunk of a working day per person, per week, that used to go on admin and now goes on actual client work or business development.
The other real win is searchability. If a client asks "didn't we discuss the budget for this back in March," you can search the transcript instead of scrolling your memory or your inbox hoping you wrote it down somewhere. That single feature has saved more awkward "let me check and get back to you" moments than anything else in my stack this year.
When to switch it off entirely
Not every call needs a bot in it. First conversations with a new prospect who's clearly nervous, calls involving anything legally sensitive like a dispute or a redundancy conversation, and calls where you want the other person to speak freely without wondering who else might read it later, all of those are better done the old way: you, a notepad, and full attention. The tool is for routine, repeatable calls, not every single conversation you have.
I've started thinking about this the way I think about automation generally, which is that the businesses that use it well are the ones who chain it deliberately rather than switching everything on and hoping for the best. The same logic that made IFTTT's whole marketing strategy work, small, specific, chosen automations rather than blanket ones, applies here. Choose which calls get the bot. Don't let the calendar choose for you.
There's also a brand angle worth thinking about if you're a consultant or coach whose whole pitch rests on trust, the way Calm built its entire brand around feeling safe and unhurried. A client who feels silently recorded doesn't feel calm, whatever your onboarding email says about "putting people first." The two things have to match.
If you're deciding whether to hand this over to someone else
I get asked a version of this question by clients most weeks now: should we just get someone in to set up the whole stack, notetakers, CRM syncing, follow-up emails, so nobody on the team has to think about consent settings or deletion rules individually. For a team of three or more people all taking client calls, that's often the right call, because getting it wrong once across a whole team is a bigger risk than getting it wrong once yourself. If that's where you are, it's worth reading through what hiring an AI consultant costs before you commit, because the range is wider than most people expect and a lot of businesses overpay for something that's a half-day job.
For a solo business or a very small team, you don't need outside help. You need the five steps above and about twenty minutes on a Tuesday afternoon.
The wider point about recording everything
There's a company that built an entire empire on the idea that people want to record their own lives and hand the footage to the world without much friction, which is more or less what GoPro's marketing strategy was built on from the start. AI notetakers are doing the quieter, business version of the same thing: everything gets captured by default, and the person being captured is the last to know it's happening. GoPro's customers chose that trade-off for themselves. Your prospects haven't, yet, unless you tell them.
None of this means don't use the tools. Use them. They save real time and they've improved how I follow up with clients. Just stop letting the default settings make a legal and ethical decision that should be yours to make on purpose, out loud, at the start of the call.
Frequently asked questions
Do I have to tell people I'm using an AI notetaker on a call?
In the UK you're processing personal data the moment you transcribe someone's voice, so under UK GDPR you need a lawful basis and the other person generally needs to know it's happening. In US states with two-party consent laws, both people legally have to agree. The safest habit, wherever you're based, is to say a single sentence at the start of every call: mention the tool, ask if that's fine, and move on.
Which AI notetaker is best for a small business?
Fathom is free with unlimited meetings and is a solid starting point for solo consultants or very small teams. Fireflies (around $10 to $19 a month per seat) has stronger search and integration with CRMs. Otter's Business plan (around $20 a seat) suits teams already inside Google Workspace. None of them are meaningfully better at transcript accuracy than the others once you're editing the summary anyway, so pick based on price and where your team already lives.
How much time does an AI notetaker save?
For a solo consultant doing five or six client calls a week, going from a full manual write-up (around 45 minutes per call) to reviewing an AI-generated summary (around 15 minutes) saves roughly three hours a week. That's the real, measurable number, not the vaguer "saves you time" claims most vendors use.
Should I turn off my notetaker for certain calls?
Yes. First conversations with nervous new prospects, anything involving a dispute or sensitive HR matter, and any call where you want someone to speak completely freely are better handled without a bot in the room. Save the AI notetaker for routine, repeatable calls where the record itself is the useful part.
Related reading: Why AI Pricing Advice Is Quietly Training Small Business Owners to Undercharge and Why I Turned Off AI Meeting Notes On My Client Calls (And What I Do Now).
I go much deeper on this in the AI marketing guide.
