Article summary
Over the past 14 months on my current project with an AI startup, I’ve typed out an average novel length’s worth of meeting notes. Here’s a quick summary:
| Metric | Value |
|---|---|
| Total files (1 file per meeting) | 132 |
| Total words | 77,967 |
| Total characters | 429,329 |
| Average words per file | ~590 |
| Average characters per file | ~3,252 |
It might seem a little crazy to write a novel’s worth of meeting notes on an AI project in a world where AI note-taking has become widespread. But I continue to find real value for myself, my teammates, and my project by taking meticulous notes in nearly all of the client meetings that I attend.
Context
I typically spend at least 5 hours a week in meetings with the client for my project. The whole Atomic team meets with the client for an hour on Tuesdays and Fridays, and I represent the Atomic team at an hour-long product collaboration session on Mondays and Wednesdays, plus we have a daily standup between a subset of the Atomic team and the client.
These meetings all cover wide-ranging topics and they’re typically all very technical. They also usually follow a round robin format where one person at a time is presenting their recent work to the rest of the team, or discussing their blockers. I’m asked to weigh in at various points in all of these meetings so it’s important for me to stay focused on the current topic and also note down action items for my team.
Value for me
One of the biggest wins I get out of taking my own meeting notes is that it forces me to stay present in the meeting. I’m a naturally fidgety person and not a great auditory learner, so it can be hard for me to stay focused on a conversation that I’m not actively involved in. I also constantly overestimate my ability to multi-task, and I’m often tempted to keep working on a story, debug an issue, or respond to Slack messages in the background during a meeting.
Without fail, the moment I start multi-tasking is the moment I’ll get asked a question that I completely missed the context for. Actively taking notes throughout the call keeps me on task by giving me a tangible goal that keeps my mind from drifting: note down what was discussed and how it may impact me or my teammates.
Value for my teammates
I’m a leader on a fairly large team, and I try to help protect the rest of the team’s heads down time by taking more of the meeting load. This is a good optimization for getting work done but the tradeoff is that the rest of the team can end up feeling out of the loop on why they’re being asked to build a certain feature, or what the client developers are working on.
My meeting notes help combat this by providing a focused list of action items or interesting updates that I can share with the rest of the team or with individuals as needed. Technically the Gemini meeting notes that get created for these calls could do that too, but they’re often much wordier than my own notes and not focused on items that impact my team specifically.
I just add a “⭐” emoji next to any items that I need to share with the team which makes it quick and easy to scan for action items. This means none of my teammates has to read through multiple pages of AI-generated notes to stay informed.
Value for my project
Meeting transcripts and AI-generated notes definitely have their uses, but I often find that they fall short when it comes to technical discussions. When Gemini is generating meeting notes, it doesn’t have the context of the product, codebase, or team members available to it. It only knows what is being said in the meeting. So it’s inevitable that it’s going to end up fudging the details, like when the team discusses the root cause for a complicated bug.
When I take notes myself, I have all that context in my head and I can take more cohesive notes because of it. I’m not transcribing what the other attendees are saying verbatim and summarizing; instead I’m picking out the important parts and rephrasing them in a way that makes sense to me as a deeply knowledgeable developer on the team. This means my notes are typically more concise and also more likely to actually be useful after the meeting when we need to remember why we implemented something a certain way, or what endpoint we should be calling when building out a feature.
Wrapping up
So yes, I’m writing tens of thousands of words a year while the rest of the industry is letting AI handle it. But I’m going to be in these meetings anyway. The only question is whether I spend that hour half-listening and half-distracted, or fully engaged with my hands on the keyboard. Taking my own notes keeps me present, gives my team a useful artifact that they wouldn’t get otherwise, and leaves me with a record that actually reflects what happened instead of what a transcript thinks happened. It costs me nothing extra and gives me more than I expected, so I have no plans to stop anytime soon.