For a while now, most of my meetings have had an extra attendee. It says nothing, promises nothing, and an hour later sends everyone a summary with a list of tasks, owners, and deadlines. Clean, fast, no typos. Exactly what a well-run project meeting is supposed to leave behind.
The first weeks with a tool like that were pure relief. No more taking notes mid-conversation, no more reconstructing from memory who said what. Then came a meeting where the summary was flawless and the project still turned the wrong way. That’s when I understood what the tool doesn’t do.
What’s genuinely solved
Credit where it’s due: recording what was said is no longer a problem. The transcript is more accurate than my notes. The task list pulled from the conversation is complete more often than mine would be, because during the meeting I’m running the conversation, while the tool is only listening.
If your pain was that nobody knew what had been agreed after a meeting, that pain really does shrink. Take the win, and don’t pretend the old pen-and-paper method was better.
What no summary will capture
Back to that meeting. We were discussing a delay to one module’s delivery. The summary read correctly: “The vendor team will deliver the module on the new date. Owner: vendor-side manager.”
What wasn’t in it? That the vendor-side manager said “we’ll try,” not “we’ll deliver.” That after he said it, there was a moment of silence and his boss looked out the window. That nobody asked what happens if “we’ll try” isn’t enough.
The tool recorded the task. It didn’t record that the task had a weak owner. Catching that was my job, and for a few weeks I didn’t do it, because I had the sense that “it’s all in the notes anyway.”
A note is a record of the conversation, not a verdict on the decision
That’s the core of it. The summary says what was agreed. It doesn’t say whether it was agreed well.
Does the task owner have the real authority to get it done? Does the deadline come from the plan, or from someone wanting to end an uncomfortable conversation? Does the decision everyone nodded through quietly push the problem onto someone who wasn’t in the room? Those questions get answered by the PM, not the transcript.
I’ve also noticed a side effect few people talk about. When notes arrive automatically, people stop reading them critically. Writing notes by hand forced me to pause for a moment: what actually matters here? Automatic notes arrive finished and look complete, so it’s easy to skim them and tick them off.
How I use it today
I didn’t turn the tool off. I changed what I do after the meeting.
Before the summary goes any further, I read the task list with one question for each item: do I believe this will happen? Wherever the answer is “I’m not sure,” I add one sentence of my own. Sometimes it’s a question for the owner, sometimes a note to myself to come back to it in a week.
The second habit: for decisions that cost something, I add to the summary who made the decision, not only who will carry it out. The tool doesn’t tell the two apart, because in conversation they sound alike. In a project they are two very different things, especially when, three months later, someone asks why we did it this way and not another.
Trust moves, it doesn’t disappear
PMs who don’t trust automated notes are right, but for a slightly different reason than they usually give. It’s not that the tool will get the record wrong. It will get it wrong less often than we do. It’s that the record was never the hardest part of the job.
The hardest part was always sensing which of the agreed things won’t happen. You can’t hand that to any tool, but you can have more time for it, now that you no longer have to take notes.
All situations described in this article are based on real events, but contain no company names, no individual names, and no data from any specific project.
