You’ve sat in that meeting. The one where forty-five minutes pass before anyone says something that couldn’t have been an email. Someone shares their screen. Someone else is clearly answering Slack messages. The meeting ends with “let’s circle back next week,” and everyone quietly knows that next week’s meeting will sound exactly the same.
Now multiply that by every team in your company, every week, for a year.
Microsoft’s Work Trend Index found that weekly meetings have tripled since 2020. That’s not a rounding error. It’s a massive shift in how companies spend their most expensive resource: people’s time. And yet, almost nobody is asking whether those hours are actually producing anything.
Most companies have no idea what bad meetings cost them
Here’s a number worth sitting with. Professionals spend roughly 392 hours a year on video calls, and about 30 percent of that time is wasted. Take a 100-person company. That’s around 12,000 hours a year burned in meetings where nothing gets decided. At $75 an hour fully loaded, you’re looking at close to $900,000 gone. Not on a failed product launch or a bad hire. On meetings.
And the real damage isn’t even the wasted time itself. It’s what happens downstream. Decisions don’t get made, so they get made later, badly, over email. Context from one meeting doesn’t carry to the next. Follow-ups slip. Deals slow down. Strategy drifts. The cost compounds, but it never shows up on a balance sheet, so nobody owns it.
What the best teams actually do
The teams that run good meetings don’t have some secret framework. They just do the boring stuff consistently.
Agendas go out before the meeting. Not a vague subject line. An actual list of what needs to be decided. People show up having read it. The conversation starts at the decision point instead of spending the first ten minutes getting everyone up to speed.
During the call, there are ground rules that people actually follow. Mute when you’re not talking. Camera on if it’s a small group. Use the hand-raise button instead of talking over each other. These sound trivial. But they’re the difference between thirty minutes that produce a clear outcome and an hour that produces a follow-up meeting.
And after? Summaries go out the same day. Action items have names and deadlines attached, not “someone should probably look into this.” Recurring meetings get killed when they stop being useful.
None of this is exciting. But the companies that do it well execute faster. Their people stay longer. And they don’t waste a million dollars a year sitting in rooms wondering why they’re there.
AI just changed the equation
Something interesting happened in the last couple of years. AI got good enough to actually help during meetings, not just after them.
The first wave of AI meeting tools was all about documentation. Record the call, spit out a transcript, generate a summary. Useful, sure. But that problem’s been solved. Every major platform has some version of it built in now.
What’s newer is AI that works in real time. You’re on a call, someone brings up a decision from three months ago, and the relevant details just appear on your screen. A client mentions a contract term, and the numbers are there before you have to say “let me get back to you on that.” It’s less note-taker, more copilot.
The AI meeting assistant market tells the story: $3.67 billion in 2024, projected to hit $72 billion by 2034. That kind of growth doesn’t come from better transcription. It comes from tools that make people sharper in the room.
The trust problem that could slow everything down
There’s a catch. An AI that listens to your meetings in real time is, by definition, always listening. And for companies in finance, legal, or healthcare, that raises questions that can’t be hand-waved away.
Where does the audio go? Who can access it? Does it train a model somewhere? These aren’t hypothetical concerns. They’re the reason most enterprise compliance teams take months to approve new tools.
The products gaining ground in regulated industries are the ones that built for this from day one. On-device audio processing, AES-256 encryption, and clear commitments that conversation data never trains models. Convo, for example, processes everything locally and stays invisible to other participants on the call.
And the regulatory picture is tightening. The EU AI Act now classifies certain workplace AI tools as high-risk and outright bans AI-driven emotion recognition at work. For any company operating across borders, privacy-first design isn’t a nice-to-have anymore. It’s table stakes.
The companies that fix this will outgrow the ones that don’t
Every company tracks revenue per employee, customer acquisition cost, time to close. Nobody tracks the quality of their meetings. But those meetings are where the actual work happens: the decisions, the alignment, the problem-solving that either moves things forward or lets them stall.
The companies that figure this out won’t just save time. They’ll make better decisions, faster, with less noise. And the ones that don’t will keep wondering why everything takes so long.
