This tech founder says prompting AI agents is a lot like managing people

The best managers of AI agents may be people who have already managed humans, one AI founder said.

Ivan Burazin, the cofounder and CEO of New York-based AI firm Daytona, has a team of about 30 people across teams in the US and Europe. He told Business Insider that in recent months, he's pushed his staff to build more AI agents; his 16 engineers each run an average of five agents, and no one at the company writes code anymore.

That's brought one thing to light for him: Staff who have managed others are way better at prompting AI agents to get exactly what they want than those with no people-management experience.

"We've found that people who manage people and understand how to communicate what they want — inputs versus outputs — they're much better at running agents," he said.

On the other hand, an engineer who's not managed anyone, even if they're great at their job, is wired like an individual contributor, Ivan said. He said they implicitly expect that people — and agents — will understand what they're asking for, and get stressed and frustrated when it doesn't pan out.

"And so they never think about how to communicate, 'This is what I want. This is how I expect you to do it, or what I think the outcome should be,'" he added.