If you run a small team, AI meeting notes are one of the easiest wins in your week. No one has to scribble during the call, no one has to write up the summary afterwards, and nothing important slips through the cracks. The question is not whether to use one of these tools. The question is which kind to use, because there are two very different approaches. The first sends a visible bot to join your call, like Otter.ai, Fireflies.ai or Fathom do. The second transcribes quietly in the background without ever appearing, which is how Granola works. The difference sounds small, but it changes who gets notified, what your guests see, and how much responsibility falls on you for consent. For most small teams, the visible bot is the safer choice for client meetings, while the quiet transcriber can work well for internal meetings, as long as you tell people it is running and write down the rule.
Two approaches to the same job
Every AI notetaker does the same core job. It listens to your meeting, turns the speech into text, and hands you back a summary with the action items. The tools just differ in how they get into the room.
Approach 1: The bot that joins your call
Tools like Otter.ai, Fireflies.ai and Fathom send a notetaker bot into your video meeting. It appears in the participant list like another attendee, often with a name that makes its purpose clear. Everyone on the call can see it, and most meeting platforms announce when a new participant joins. If a guest is uncomfortable, they can ask for it to be removed or leave the call. This approach has real momentum behind it. Otter says its platform now has 35 million users, according to a TechCrunch report, and Otter’s chief executive told TechCrunch that most of its enterprise customers actually prefer the bot that joins the call, because it provides transparency.

Approach 2: The app that listens quietly
Granola takes the opposite path. Instead of joining the call, the app runs on one person’s computer and captures the audio from that device, then blends the transcript with whatever notes that person typed during the meeting. No bot appears on screen, and no other participant is told that a transcript is being made. Investor interest shows where this category is heading. Granola raised 125 million dollars in a Series C funding round at a 1.5 billion dollar valuation, led by Index Ventures, according to a TechCrunch report. The quiet design is the product’s whole appeal: meetings feel normal, and the notes appear afterwards as if by magic. The trade-off is that the invisibility is the feature and the risk at the same time. Something to note is that the two approaches are starting to converge. Even Otter has added a bot-free capture mode for its desktop apps, according to a TechCrunch report, so the choice is increasingly about defaults and settings rather than brand names.
The four tools side by side
Here is how the four best-known tools compare on the details that matter for a small team: how they join your meetings, what the free plan gives you, and where paid plans start. Prices are per user per month in US dollars, checked against each vendor’s pricing page.
| Tool | How it joins | Free plan | Paid plans from | Best for |
|---|---|---|---|---|
| Otter.ai | Bot joins the call | 300 minutes a month, 30 minutes per conversation | $8.33 annual billing ($16.99 monthly) | Solo users who want the cheapest reliable plan |
| Fireflies.ai | Bot joins the call | Unlimited transcription, limited AI summaries | $10 annual billing ($18 monthly) | Teams wanting unlimited transcription |
| Fathom | Bot joins the call | Unlimited recording, 5 AI summaries a month | $15 annual billing ($19 to $20 monthly) | Sales teams needing CRM sync |
| Granola | No bot, captures quietly | Basic plan with capped history | $14 (Business plan) | Teams comfortable with invisible capture |
What guests actually see, and why it matters
Think about this from the perspective of a client or a job candidate who joins your call. In the first approach, they see a notetaker in the attendee list, they understand what it is doing, and they can object. In the second approach, they see nothing. They speak freely, assuming the conversation stays in the room, while a transcript is being written on someone else’s laptop.
For internal team meetings, most people will accept the quiet approach once they know the rule. For client meetings, interviews, or any conversation with people outside your company, the visible bot is almost always the better choice. A small business runs on trust, and trust is easier to keep when every participant can see who, or what, is in the room. Announcing the tool at the start of the call also protects you if a guest later claims they did not know they were being recorded.
The privacy catch: consent is your job
Here is the part most small teams skip. A tool that guests cannot see does not remove the consent question. It moves it onto you. The rules on recording conversations change depending on where you are. In the United States, some states allow recording when only one participant consents, while others require every participant to agree, according to legal analysis in Corporate Compliance Insights. Outside the United States the rules differ again, and they can be stricter. This is not legal advice. It is a practical warning to check the rules for your own state or country before you switch any tool on.
The quiet approach has already drawn legal attention. A proposed class action filed in a federal court in California alleges that Granola captured meeting audio without other participants knowing, according to legal analysis in Corporate Compliance Insights. The allegations are unproven, and the case is still at an early stage, but the underlying point is worth hearing. The legal risk sits with the company that deploys the tool, not just with the vendor that built it. Otter has faced similar lawsuits over consent, according to the same analysis, including a court decision that refused to dismiss several of the privacy claims. A visible bot does not automatically solve the consent question either. What protects you is a habit of disclosure, not a particular product design.

The practical takeaway is simple. Tell people the tool is running, every time. Put it in the calendar invite, say it at the start of the call, and give guests a clear way to opt out. If you use the quiet approach, that disclosure matters even more, precisely because the tool will not do it for you. Also check where the tool stores its data, how long it keeps it, and whether it uses meeting content to improve its own models, then turn off anything that does not fit your business.
How to choose for a small team
Work through these five questions before you commit, and the right tool will usually pick itself.
1. Who is in your meetings? If most of your meetings involve clients, candidates or partners, choose the visible bot. If your meetings are mostly internal and everyone has agreed to the setup, the quiet approach can be a good fit.
2. Who should see the notes? Some teams want every attendee to get the summary automatically, which the bot-style tools handle naturally. If the notes are mainly for your own follow-up, a personal tool is fine.
3. How will you handle consent? If your team will not reliably announce the tool at the start of each call, the visible bot is doing part of that work for you. The quiet approach only works with a team that discloses it every single time.
4. What happens to the transcript afterwards? Read the data and privacy settings before you roll anything out. Check where transcripts are stored, how long they are kept, whether meeting content can be used to train models, and whether you can set a retention policy. These settings matter more than the transcription quality for most small teams.
5. Can you trial it on one meeting first? Pick one regular weekly meeting, run the tool for a month, and ask everyone how it felt. That feedback will tell you more than any feature comparison.
A simple policy your whole team can follow
Once you choose a tool, write down a short rule so the whole team follows it the same way. It does not need to be a legal document. Three lines are enough: we tell every guest the notetaker is running, we note it in the calendar invite, and we delete transcripts we no longer need. If writing policies feels unfamiliar, our guide to a simple AI policy for your business walks through the same idea in more detail. The teams that get into trouble with these tools are rarely the ones with the wrong product. They are the ones with no agreed habit at all.
Frequently Asked Questions
Which approach fits a small team best?
For most small teams, the visible bot is the safer default, especially for meetings with clients, partners or candidates. It tells every participant that a transcript is being made, which builds trust and covers you on consent. The quiet approach can work well for internal meetings where the whole team has agreed to it in advance.
Do we have to tell guests that an AI tool is recording the meeting?
You should always tell them, whether or not the law in your area requires it. Put the note in the calendar invite, mention it at the start of the call, and let guests opt out if they prefer. Disclosure is the habit that protects your business, regardless of which tool you use.
What is the real difference between Otter.ai and Granola?
Otter.ai joins your meeting as a visible participant that everyone can see, while Granola captures audio from one person’s computer without appearing on the call. Both produce transcripts and summaries, but guests will know about the first one and will not know about the second one unless you tell them.
Can these tools use our meeting data to train their AI models?
It depends on the tool and the settings you choose. Some tools include settings that allow meeting content to be used for improving their models, so you should read the privacy settings before rolling the tool out to your team and switch off anything that does not fit your business.
Is a visible bot always the more transparent option?
It is usually the more transparent option, but it is not a complete answer on its own. A bot appearing in the attendee list does not automatically mean every guest understood what it was doing, and courts have not treated its mere appearance as consent. The safest position is a visible tool plus a clear spoken announcement at the start of the call.
How do we start a simple meeting-recording policy?
Start with three commitments that the whole team can remember. Tell every guest the notetaker is running, note it in the calendar invite, and delete transcripts you no longer need. Write those lines down where your team will see them, and review them once a year or whenever you change tools.
Meeting notes are a natural first step into using AI in your daily operations. If you want help turning those notes into real follow-through, read our guide to automating meeting follow-ups and the business tasks to automate first. And if your team needs a calm, practical introduction to working with AI tools, our AI training for your team covers the habits that make tools like these actually stick.
Want help putting this into practice?
If you’d like guidance setting up AI tools for your team, book a call with us. We’ll look at your workflows and tell you honestly what’s worth automating.


