AI tools for email, meetings, and writing and how they work


There are hundreds of thousands of AI tools available for workplace tasks, with more coming on the market everyday. Automation and AI is already changing hiring and recruiting and will eventually impact everyone’s job in some way. So on the latest episode of The New Way We Work, we wanted to see if a few of these AI workplace tools could make our jobs easier, removing some of the more tedious tasks that come with most white collar work and free up more time. The results were mixed.


I tried out MeetGeek for a week. Once I synced it to my work calendar, the bot would automatically join every Zoom meeting I was in. (There are also integrations with Google Meet and Microsoft Teams, but not Slack where I use huddles for many of my more informal check-ins.) MeetGeek let participants know it was listening and taking notes and automatically emailed notes and a summary to everyone at the end. The summary included “concerns” and “actions,” organized everything that was discussed into categories in a “topics and highlights” section and included “AI Insights,” which analyzed the sentiment, productivity, and other highly subjective observations. 

Those insights were what I and others who received the note found most uncomfortable. My weekly check-ins with senior editor Christopher Zara are congenial and casual, as we talk about both the practical aspects of our work and the news and story ideas. But the meeting bot indicated that I was often “skeptical” and “concerned” and even though we mentioned several story ideas and things to follow up on, the notes concluded that we had no further “actions.” Zara remarked that seeing summaries like this sent to his colleagues would make him more self-conscious about speaking up in meetings.

On the positive side, Meetgeek not only works for meetings that you aren’t hosting, but it even works on meetings that you are invited to but don’t attend. Later in the week, I had to miss a leadership meeting and it was helpful to return to my desk to find comprehensive notes (though I did still check in with my colleagues to see what they deemed the most important takeaways). 


Senior staff editor Max Ufberg used Sanebox for a week. The email AI program sorts all of a user’s emails into various folders like “news” and “receipts.” “[It was] nice to wake up in the morning and see a clear inbox because everything has been sorted into subfolders,” Ufberg said, with the caveat that “this product, like many AI tools, gets better over time. It learns your habits and your preferences.” So using it for a week didn’t make his workday easier, he said, as he had to go through each of the folders to make sure he didn’t miss anything important. “It wasn’t particularly accurate. I would have important news put in the spam folder and vice versa,” he said. “If there’s one item that you think might be miscategorized, then you effectively have to look at every item.”

Writing and idea generation

Podcast producer Julia Shu used ​​Swell AI, a tool for podcast production and Um, and Descript for audio transcription. Fast Company’s parent company Mansueto Ventures has an AI policy that no published work may be created by AI, so Shu used Swell AI just as an exercise to compare how our human-written podcast episode headlines and descriptions compared to those generated by AI.

Shu found the AI-generated ideas to be straightforward and academic. “A lot of them sounded like college papers or peer-reviewed journal articles,” she said. “There’s a lot of colons, which is something I would do when I was in high school and I had to write an essay.” In other words, not the kind of writing human journalists use to entice human readers to want to click, read, or listen.

Shu did find tools like Descript useful, however, as the podcast team has been using it for years and the linked transcripts and audio have gotten more accurate and made editing easier. 

Bottom line

As Ufberg pointed out, all AI tools like these have to be trained to learn what you deem important and what your preferences are, and a week isn’t enough time to train AI. As he said, it’s a long-term commitment—one that often comes with a monthly subscription fee.

There’s also the chance that the program you select might not last. “There [are] still like 30 programs competing for every task right now,” Ufberg explained. “We’re still in such early stages. You’re kind of betting on a horse and hoping that the horse doesn’t drop out early.”

Listen to the full episode for more on these AI tools and they impacted our daily work.


Source link

Be the first to comment

Leave a Reply

Your email address will not be published.