Two emails summed up the situation with Peter Choi. One was from Peter himself, which began by asking if I was “working hard, or hardly working?” and then sought permission to attend a conference in Denver, which would give him “an excuse to break out my ski gear.” Peter was a longtime, trusted employee, but lately the joking seemed to mask a sharp decline in his performance. The second email came from a client complaining about some shoddy work Peter had turned in. As Peter’s manager, I would have to address both at today’s check-in and make a decision: either send him to the conference, with or without conditions, or not—and maybe even propose a formal intervention.

And that’s when I pushed a button to begin the interaction—Peter is a bot developed by the corporate training company Abilitie. Powered by artificial intelligence, designed to calibrate his responses based on what I say to him, Peter Choi is a character in a new series of what Abilitie calls “case studies” for teaching business skills.

Corporate trainers increasingly see AI as the future of interpersonal skills development, says Whitney Wirta, director of enterprise learning at the Hanover Insurance Group, which uses Abilitie’s AI cases in a training program for its new managers. “We can throw knowledge at people,” says Wirta, “but they really need a safe space to practice, take those at bats—to try something, get feedback and try it again.”

Abilitie has been making business simulations, in which players chart a company’s path over several quarters, since 2015, and Chief Executive Officer Bjorn Billhardt has been doing this work since 2001. He, like so many of us, was caught off guard by the sudden emergence of AI in late 2022, but once he had his fun asking a chatbot to compare pizza to hamburger, he began to think about how the technology could transform his work. “For 20 years, my bread and butter has been to put people in scenarios where they feel the pain of making difficult decisions and then see the consequences of their actions,” he says. “And we’ve devised all sorts of innovative ideas”—mainly variations on multiple-choice questions—“to get around the fact that the computer is, in fact, not a human.”

The company’s first move, in mid-2023, was to incorporate the technology into two of its simulations, as interactive characters that serve as complications or plot points in the unfolding narratives. It eventually became clear, Billhardt says, that AI was especially suited to animating the discrete situations and moments of decision often analyzed in business case studies. “The business school case study is a proven methodology that hasn’t been in any meaningful way reimagined for decades,” he argues. But AI, he says, can transform the case study into a multilayered experience where, for example, a user may talk with several characters giving conflicting information. “You have to first decide, which character do you believe? Then you have to decide, how do you share your decision with all three of those characters? Oftentimes the hardest part of any business case is not making the decision; it’s how you frame it right and communicate it.”

The result, after eight months of work and about $2 million, is a collection of 24 separate cases, each an interaction with employees, colleagues, or interested outsiders designed to last about 20 to 30 minutes. (Abilitie is packaging the cases with facilitation and debriefing, a program that in total should run from one to two hours.) The case subjects are evenly divided between people management and financial topics, but even the latter are heavily dosed with soft skills practice in listening, explaining and persuasion.

If my conversation with Peter didn’t feel fully authentic, it did somehow compel me to fully engage in it and take it seriously. That’s by design. For instance, Peter continually asked me for advice. “Pushing a learner to articulate things engrains the core concept that they’re exploring,” says Luke Owings, Abilitie’s vice president of product. And certain prompts from the user will bring out changes in the character. When I told Peter that his colleagues found the horseplay distracting, that, Owings told me, was his cue to become more serious, though I didn't detect the change.

“The machine does some amazing things,” Owings says. “It is an earnest and tireless role-play provider. And it gives a learner an immediate understanding of what they did well.” In one exercise, the machine dinged me for not being specific enough in establishing deliverables; with Peter, I set firm deadlines. It also made plain that sometimes I hadn’t been as explicit or clear with my counterparties as I had thought. (For facilitators managing a class, it also combs through all of the interactions and highlights the best exchanges for a discussion.)

“It’s quite robust feedback, it’s pretty targeted,” says Hanover’s Wirta. So far, 75 managers at the group have used Abilitie’s AI cases—now 12 cases over the eight-month program. “The workflow was tight. They set the stage quite well for the program, giving you background about the person you’re dealing with and the outcome of the conversation you’d like to achieve.”

When Wirta began developing the training program late last year, few other companies offered off-the-shelf simulated case studies like these. He expects that will change as more people recognize the opportunity. For his part, he hopes to deliver case studies to more employees. “We’ve found that people in the cohorts liked it so much that they shared it with their managers, and their managers have reached out to us about how they could use it for other members of their team more broadly,” he says. “We’re looking into how we could expand it for next year.”