I tested Read.ai, Fireflies and Otter on my real meetings. Only one stayed.
An AI note-taker is only worth paying for if it hands you notes you can use. A transcript on its own doesn't do that. In a test I posted in March 2023, I ran Read.ai, Fireflies and Otter on my real meetings, and only one of them stayed.
The business case
My day is back-to-back calls, and a week or two later I need to understand what was discussed in each one. Writing that record myself takes time I don't have between meetings. At 10 bucks a month back then, an AI assistant that writes it for me was worth it.
In the video I also asked for the step after the notes: give me some more programmatic tools so I can massage the notes a little better.
I joked that the bots are going to take over our lives anyway. The future I described is more practical. After each meeting, a webhook triggers one of my internal services, the notes go into my own note system, and an email goes out to everyone who was on the call.
The implication I took from the test is that a note-taker has to do better than a transcript. If I still have to copy the notes into Notepad piece by piece before I can send them, the tool is so close and still falls short.
My notes had no structure
I jump between meetings all day. The day I recorded this, I had a 7:30, then an 8:00, and I squeezed the video in with another call starting 10 minutes later.
There's no time in between to organize anything.
My system was a OneNote notebook with a preset layout. Tuesday meeting notes, Wednesday meeting notes, and so on. In practice I ended up writing with my iPad pencil instead.
Handwriting helps me retain the information, so it's fine for that. The problem is that it isn't structured, and I can't do anything else with it.
I figured I'd let the bots fight it out and see which one was right for me.
Where Otter and Fireflies landed
I was on trial plans for both in March 2023. Paid plans and today's versions may look different, so treat this as what I saw back then.
Otter was jumping into every meeting, and it did a good job of transcribing them. It gave me a summary and the transcript. I went looking for action items, comments and highlights, didn't see any, and kept getting prompts to upgrade. For what I needed, it wasn't worth keeping, so I closed the account right there on the recording.
Fireflies showed my past meetings and the ones coming up, and it picked up a meeting title that Read.ai missed. On the trial, though, the transcript sat behind an upgrade, while the other tools gave me transcription by default. Even then, transcription seemed to be the main feature.
I walk through all three tools on screen in the full video on my channel: Read.ai vs Fireflies vs Otter.ai on YouTube.
Why Read.ai stayed
Read.ai was the one I kept, and at the time it was the best one I'd seen. Otter and Fireflies felt like the same product to me, and Read.ai had the best set of features. These are the parts I walked through on screen:
- A meeting summary that boiled the whole transcript down to what was discussed.
- A breakdown of who talked the most, and, no surprise, in that meeting it was me.
- A meeting score and engagement numbers, which I called fluff but kind of cool.
- The full transcript, plus the topics and the action items.
- Automatic joining for Teams and Zoom calls, because I'd connected it to my Google Calendar.
I appreciated the summary most. That's the thing my handwritten notes never gave me.
What I still wanted
Read.ai wasn't perfect, and I said so in the video.
Sharing was the big one. Anyone I sent a meeting link to had to log in to see it. I know why they do it. It brings in signups, but it drives me nuts. I want to send notes to someone without making them sign in.
Exporting was the second. In a perfect world it would give me a clean PDF or a plain document. Instead I had to copy the notes into Notepad section by section, and they lost their structure on the way.
The email it sends to participants was better, but the formatting still had a lot of extra stuff around the notes. I wanted to drop the summary into an email, make a few edits and be done.
The third was a webhook, so Read.ai could trigger one of my internal services after each meeting and I could shape the notes myself.
Judge it by the output
I care about what I get after the call. I want a summary I can read, action items I can act on, and notes I can send without rebuilding them.
Test a note-taker on your real meetings and look hard at what it hands you afterwards. If you still have to clean up the notes yourself, it isn't saving you much time.
The point of AI notes was never the transcript. The point is walking out of back-to-back meetings with a record you can still use two weeks later.
If you already have customers and want AI and simple systems to take more admin off your plate, that's what I cover in The Scaling Entrepreneur.