CEOs have a new skill to learn: managing AI employees

Sarah Franklin, the CEO of Lattice, previously tried giving AI workers employment records.

Carlos Rodrigues/Sportsfile for Web Summit via Getty Images

Agentic AI, which makes decisions without human input, is a hot topic at Davos this year.AI agents promise tangible benefits to companies but raise questions such as whether they should get KPIs.Business leaders now have to work out how to manage autonomous bots alongside human employees.

Many companies are using AI to get better, faster, and leaner — but what happens when artificial intelligence is not just your tool but your colleague or your employee? Should it be trained like a human, given goals and performance metrics?

There’s a reality setting in at the World Economic Forum in Davos this year: business leaders will soon have to decide how to manage not only their human workforce but also a new class of AI employees while maintaining harmony between bots and people.

It’s something Lattice CEO Sarah Franklin has spent a lot of time thinking about. Her company, which builds HR software, announced last year that it was going to treat AI workers like humans by giving them official employee records and even putting them through onboarding like a real human employee. The actual humans weren’t keen on the idea and Lattice hit pause after the backlash.

Now, with talk of AI agents everywhere, Franklin says Lattice’s idea was “really prescient” if too early. “We were ahead, but by months.”

That may prove to be so. Agentic AI is a hot topic at Davos this year. The past two years have brought much hype around artificial intelligence, but agents — AI that can act and make decisions independent of user input — are what many stakeholders believe can provide tangible, immediate benefits. Or, as one exec in Davos put it, the time for kicking the tires is over. Businesses want returns on their investments — and agents are one way to get it.

Franklin is not the only one raising the flag on this topic. “I don’t think the world has yet had the opportunity to think through all of the implications,” Alan Flower, global head of AI at HCLTech, told Business Insider.

“For example, as a manager, I’m going to be managing a human workforce and an agentic workforce at the same time — they’re going to have to collaborate,” said Flower. “My agents are going to have to collaborate with agents from another company, for example.”

The question then, said Flower, is how do employers “broker” collaboration between agents — including from other companies —and motivate them to do good work?

“Will we get to a phase where agentic AI will be given KPIs? These are all considerations that the world of work is going to have to kind of contemplate,” Flower added, referring to key performance indicators.

‘How do I manage them?’

Franklin says one of the mistakes Lattice made with its announcement last year was in how it communicated it. “People are more comfortable with the word ‘agent’ than they are with ‘digital worker,'” she said. Franklin added, “There’s a huge lack of education of what AI is and a disbelief that AI is going to be able to have the types of conversations that it can.”

That’s a huge opportunity for HR leaders, Franklin said. And, of course, for Lattice, which counts OpenAI, Perplexity, and Anthropic among its customers.

Becky Frankiewicz, chief commercial officer at ManpowerGroup, told BI she has been hearing a lot of talk about whether companies will need “AI managers” for their AI agents.

“I think the first step in working out how to manage these AI employees is, how do I deploy them and make them productive? And then step two is, how do I manage them?”

Frankiewicz said she knew one consulting company already grappling with this. “They’ve done the agents already,” she said. “The next question they were asking was: do we need to have managers for the agents?”

Lattice’s Franklin thinks the discussion around AI in the workplace isn’t being taken seriously enough — and needs to happen soon.

“The reality is that today, AI is being implemented as agents to have conversations on behalf of brands, on behalf of people,” Franklin said, adding that it’s a huge amount of trust to put into AI without proper governance. As agents take on more responsibilities, more governance will be needed — but that means acknowledging the new reality and having those conversations.

“We need everyone to get to this place where they’re comfortable, they’re not afraid,” she said. “And we can work together with AI in the workforce in a way that feels natural.”