Latest / Future of Work Tech with Fexingo: Remote Tools, AI Productivity, and Workplace Software / How AI Is Automating Your Meeting Notes
Transcript
- Lucas: Luna, I want to talk about something that every single knowledge worker deals with, but we rarely scrutinize: the humble meeting note. Luna: I have a feeling you're about to tell me that robots are taking over that too. Lucas: Exactly. And it's happening faster than most people realize. A Gartner survey from earlier this year found that 57 percent of knowledge workers now use some form of ai powered note-taking in their meetings. That's up from about 22 percent just two years ago. Luna: That's a massive jump. And I'm guessing the tools aren't just transcribing anymore. Lucas: Right. The first generation—think Otter.ai, Fireflies.ai—they were basically glorified voice to text. You'd get a transcript with timestamps, maybe a crude summary. But the new wave is doing something different. These tools are now generating action items, assigning owners, even flagging sentiment shifts in the conversation. Luna: Sentiment shifts? Like, 'Luna sounded frustrated when she said the budget was too low'? Lucas: Exactly that. Some platforms use natural language processing to detect tone—positive, negative, neutral—and map it across the meeting timeline. The idea is to give you a 'temperature check' of the room, even if you weren't there. Luna: That's impressive, but it also creeps me out a little. I mean, do we really want our frustration levels being algorithmically catalogued? Lucas: That's one of the big debates we need to unpack. But first, let me ground this in a concrete example. I spoke with a marketing agency in Austin—about 80 people—that adopted an AI note-taker across all client calls. Their claim: they reduced the time spent writing meeting recaps by 80 percent. Account managers went from spending two hours per client per week on notes to about 24 minutes. Luna: That's a huge efficiency gain. But I'd want to know: did the quality of the recaps suffer? And did the team actually trust the AI's action items? Lucas: Great questions. The agency told me the recaps were 'good enough' for routine status meetings—client updates, project check-ins. But for complex strategic discussions, they still had a human review and edit. The interesting unintended consequence: they noticed a 15 percent drop in spontaneous creative ideas during brainstorming sessions. Luna: Wait, really? Why would note-taking affect creativity? Lucas: The hypothesis is that when people know every word is being captured and analyzed, they self-censor. They're less likely to throw out half-baked ideas or make offhand comments that might later be flagged as 'negative sentiment.' The AI creates an invisible auditor in the room. Luna: So the tool that's supposed to free up mental bandwidth might actually be constraining the very thing that makes meetings valuable: the messy, generative exchange of ideas. Lucas: That's exactly the tension. And it's not just creativity. There's also a privacy dimension. Some companies have outright banned external AI note-takers because they're worried about proprietary information being processed on third-party servers. Luna: I've heard that. Especially in legal and healthcare, where confidentiality is paramount. But even in less regulated industries, there's a growing unease about who owns that data. Lucas: Right. And the terms of service vary wildly. Some tools claim ownership of the transcripts to train their models; others promise that your data is siloed. It's a mess of fine print that most users never read. Luna: So what's the path forward? Do we just accept that some spontaneity and privacy will be lost for the sake of efficiency? Lucas: I don't think it has to be binary. The smartest implementations I've seen are opt-in at the meeting level, not always-on. And they give participants the ability to pause recording during off-the-record segments. That way, you get the efficiency when you need it, but you preserve space for candid conversation. Luna: That feels like a reasonable compromise. But it also requires a culture of trust—where team members actually feel safe hitting that pause button without looking like they're hiding something. Lucas: Exactly. The technology is only half the equation. The other half is how we design the norms around it. And that's where I think the conversation gets really interesting. Luna: Speaking of things that require thoughtful design, I want to take a quick moment here. You know, we do these episodes without any ads—no sponsors, no mid-roll interruptions. It's a deliberate choice because we think this kind of tech conversation works better without commercial breaks. Lucas: It does. And we hear from listeners all the time who appreciate that. If that model works for you and you'd like to support it, there's a simple way to do it: buy me a coffee dot com slash fexingo. That's it, no pressure at all. Luna: Totally. It's how we keep this thing independent and ad-free. Okay, back to meeting notes. Lucas: So, another trend I'm watching is the integration of these note-taking tools with the rest of the productivity stack. Otter recently announced a native integration with Asana, where action items from a meeting automatically become tasks with due dates and assignees. Luna: That's a logical next step. But does it actually work reliably? I've seen demos where the AI misattributes 'Lucas will send the deck' to 'Luna will send the deck'—which can cause real confusion. Lucas: It happens. The current error rate for speaker identification is around 5 to 8 percent in ideal conditions—good microphone, clear speech. In a real conference room with crosstalk, it can be much higher. So human review is still essential. Luna: So we're not at full automation yet. What's the timeline you're hearing from the industry? Lucas: Most product leaders I've talked to say we're two to three years away from truly reliable, hands-off note-taking. The breakthrough will come when these models can handle multiple overlapping speakers, different accents, and industry jargon without error. Luna: That's a tall order. But given the pace of improvement, I wouldn't bet against it. The question is whether, by then, we'll have redesigned our meeting culture to actually benefit from all this data. Lucas: That's the million-dollar question. We have all this new capability, but the default behavior is still to have too many meetings, too long, with too many people. AI notes can make a bad meeting more efficient, but they can't make it a good meeting. Luna: So maybe the real productivity hack isn't better note-taking. It's having fewer meetings worth taking notes on. Lucas: I think you just summarized the entire episode in one sentence. And I don't have a better closing thought than that. Luna: Let's leave it there, then. Thanks for listening, everyone.