AI agents are supposed to eliminate busywork. For AI strategist Sol Rashidi, it ended up creating more work instead.
“I just fired half my agents because they were unreliable,” Rashidi, chief strategy officer at data security firm Cyera and a senior fellow at Harvard Kennedy School, told Business Insider.
Rather than freeing up her time, she said the agents — virtual assistants designed to complete tasks autonomously — demanded constant supervision.
“I was spending more time babysitting them” than doing useful work, she said. Rashidi had four agents running at one time and got rid of two.
She is one of a growing number of what a recent Glean report called “botsitters” — workers who spend hours each week feeding AI context, debugging mistakes, and cleaning up errors.
The report found white-collar workers spend an average of 6.4 hours a week on that often-overlooked work, nearly a full working day over the course of a week.
While many companies, including Microsoft, are still figuring out how their AI agents will work with their human employees, several solo entrepreneurs are running their businesses, in part with AI agents.
Still, for Rashidi, deploying AI agents in real-world work proved far messier than the hype suggests.
“I think that the hardest part is the over-glamorizing of bots, agents,” she said. “Yes, it’s amazing, but there’s a place and time for everything. And we’ve lost the ability to discern that place and time.”
‘I don’t have the time to babysit agents’
The problem Rashidi described fits into a broader “productivity paradox” that some companies are facing.
As part of “The Great Coding Reset” series, Business Insider’s Juliana Kaplan and Jacob Zinkula reported that while AI is helping many employees complete tasks faster, those gains aren’t consistently translating into better company performance.
The Glean report said that much of this missing productivity may be getting eaten up by botsitting.
That tracks with Rashidi’s experience.
Rather than continuing to manage agents, Rashidi told Business Insider she hired human virtual assistants to handle some of the work instead.
“I don’t have the time to babysit agents and keep course correcting the context,” she said.
The lesson for her is that leaders shouldn’t automate for the sake of automation.
“Individuals are going to have to apply judgment and critical thinking,” she said. “And leaders are going to have to get past the narrative of ‘we must do AI at all costs’ because sometimes the cost is higher than the reward.”
Read the full article here
