Getty Images; Alyssa Powell/BI
Matt Martin knows about meetings run amok.
He's the CEO and cofounder of Clockwise, which aims to help people manage their work calendars so they have more time to get things done — and not just sit in meetings.
Earlier this year, in a bid to be more efficient, he started using an artificial-intelligence tool called Granola to help him take notes in meetings and summarize takeaways and to-dos.
The result for Martin is time saved and "actually pretty damn good notes," he told Business Insider.
Efforts to reduce the sting of meetings are perhaps as old as meetings themselves. Yet the imperative can feel more urgent thanks to our propensity, hardened during the pandemic, to wedge more gatherings into our calendars.
Now, thanks to AI, we might soon have fewer work meetings — or at least attend fewer. It's likely, according to execs leading the development of the technology, that corpulent calendars will be no match for AI-powered notetaking apps capable of being everywhere all at once.
And AI meeting bots won't serve just as digital scribes. They'll resemble all-knowing, indefatigable assistants able to take on tasks like answering questions on our behalf, interviewing job candidates, and training workers, execs told BI.
Sam Liang generally has as many as 40 meetings a week.
It's not practical for him to attend each one, so sometimes he sends an AI stand-in. This is easy enough for Liang since he's the CEO and cofounder of Otter, an app that records audio from meetings and produces a real-time transcript using AI.
Liang told BI he uses Otter to forgo some meetings. He then reads the summaries or listens to the recording. Liang expects more leaders will soon do this.
He estimated that perhaps 20% of C-level executives would use AI avatars to attend routine meetings on their behalf by the end of 2025.
In his case, Liang has an avatar that acts like a "personalized agent." Otter trained the AI using seven years' worth of Liang's meetings, along with emails and some Google docs he wrote on topics like product principles, Otter's strategies, and why the company does certain things.
"When people ask me those questions, my avatar can answer probably 90% of those," Liang said.
This knowledge can flow to new hires at Otter. Liang said his AI avatar could use what he's said and written to explain his vision for the company, its strategies, and its origin, for example.
The ramifications of having an ever-present AI available to document our workdays — and beyond — will be similar in scale to that of the introduction of the internet, said Terry Sejnowski, a distinguished professor at the University of California, San Diego, who's a neuroscientist and the author of the book "ChatGPT and the Future of AI."
"Nobody predicted the impact it was going to have on our lives," Sejnowski said. "Same thing here. It's going to take decades."
He said keeping track of meetings and other interactions would go well beyond capturing audio or video. Sejnowski sits on the scientific advisory committee for Softeye, a startup developing glasses intended to work with a smartphone to serve as an AI assistant. Similar attempts have been made, of course. Remember Google Glass?
Ray-Ban Meta glasses allow users to take photos and videos. In September, Meta CEO Mark Zuckerberg said updates to the glasses aimed to let users translate certain languages, scan QR codes, and capture images of what they've seen so they can refer to them later when, for example, they need to buy something.
Softeye's plan, Sejnowski said, is to have glasses that constantly recognize objects and people around the wearer and provide related information. He said they would also take snapshots and store them, along with the time they were taken. That would make it possible, he said, to reconstruct where a user was — and rely on the AI assistant.
"You can ask it questions," Sejnowski said. "Did I promise anything to this person?"
Richard White, like so many other desk workers, found himself stuck on endless Zoom calls during the pandemic.
He found it frustrating to take notes, jump to another call, and have little time to clean up his takeaways in between. Plus, White said, even good notes weren't always reliable after too long.
"Do you really remember what was important?" he said.
Four years ago, White started Fathom, a company that uses AI to capture video and generate notes from meetings.
People don't necessarily want a transcript, he said, though it's often necessary for AI to work its meeting magic — including generating notes, making to-do lists, and updating data on customer-relationship management.
White said that what most meeting-goers are after, aside from a list of action items, is a better recall of the ephemeral and unstructured information that's often delivered at these gatherings. Showing up, White said, is often the only way to access it.
He said AI notetakers would be able to produce highlight reels of key meeting moments. The goal, White said, would be to reduce "meeting inflation" by enabling fewer people to attend them while maintaining information flow.
"You'll have an AI that actually goes out and listens to every meeting in your org and comes back and tells you, 'Here's the five minutes of content you should pay attention to today,'" White said.
White said an accessible record of all but the most sensitive meetings within an organization could serve as a basis for identifying gaps in training or generating feedback. That's in part because AI can now accurately discern sentiment and tone — something that's become possible only in the past six months to a year, he said.
Beyond that, he said, AI meeting bots will be able to act on ideas. So if someone in a meeting proposes creating a document, the AI would have a draft ready soon after.
White doesn't expect we'll necessarily each have individual bots that go to meetings on our behalf. He said that would quickly result in meetings swimming with AI avatars.
The best approach, White said, would be to use a "federated" system where all the meetings are accessible. That way, anyone not in the meeting could access the content through a personal agent that lives in the cloud, he said.
White said bosses could ask AI for instances in which a meeting was positive or when participants grew frustrated. A search might take the form of, "Give me a pricing discussion that didn't go well," he said. That goes well beyond parsing a transcript for the word "price," he added.
"The tech is finally there, and it's really good," he said.
AI could also help document meetings with prospective employees, said Alan Price, the global head of talent acquisition at Deel, a global human resources company that helps employers hire abroad. Price told BI that Deel had begun using AI meeting tools to reduce the time and personnel needed to hire for roles like customer service.
That's important because when Deel posts that type of job, Price said, the company might soon have some 4,000 applications. So Deel uses an AI bot to conduct an initial interview with promising candidates. Then, a recruiter can evaluate the summary of the interview and, if necessary, review the audio and video to determine whether the candidate should move on to an interview with a person.
Price said that rather than spending 30 minutes on a single interview, a recruiter could review five or six interview summaries in that same time.
That bump in efficiency has enabled a single recruiter to hire 30 to 35 candidates within about two weeks, he said.
"The recruiter makes the decision," Price said, "but it's streamlined."
