We've sometimes wished we could have our own Wendy Rhodes, the performance coach at the hedge fund on the TV show “Billions.” Most workplaces, however, aren't bringing in billions and can't afford a Wendy. But an upskilling platform called Multiverse uses artificial intelligence to provide personalized, on-the-job guidance.

Its AI coach, Atlas, helps workers expand their abilities and keep themselves relevant in an economy that makes skills obsolete faster than ever before, says Ujjwal Singh, chief product and technology officer at Multiverse. The following is an edited transcript of his conversation with Marketplace’s Meghan McCarty Carino.

Ujjwal Singh: Our focus areas across training and upskilling and reskilling include things like AI data skills, and then also moving into things like leadership skills. One of the things we pride ourselves on is we're not a traditional learning platform in the sense of we don't hand employees a library card. We're actually focused on driving real [return on investment] for companies as well as employees. And so Multiverse learners have driven over $2 billion in ROI for their employers using project-based on-the-job learning. And this is one of the key places where Atlas, our AI coach, has actually been really instrumental.

Meghan McCarty Carino: So you rolled out this AI coach about a year and a half ago, and you have some research about how that first year went. What did you find?

Singh: Yeah, so over over the last year or so, Atlas itself has received more than a half a million messages, and it has seen 4x user growth in the last six months alone. And it's being used for everything from how to calculate a return on investment for a project to explanations of technical concepts, getting clarity on where the errors are coming from in data analysis or coding. One of the things that's really interesting is that it has a 97% helpfulness rating as of this last January. And over the last year, it's been consistently above 95%. And the reason that's important is obviously just a chatbot for the sake of chatbot isn't useful. It's something that ultimately we're building to get the best outcomes for the employees and then, therefore, for the employers. So the thing that Atlas is actually doing really well is, and the thing that we can't do with just human coaching, is it's available 24/7 on the learning platform, and that's actually been really critical. And what we're finding is — going back in history a little bit — in the 1980s there was a very famous study done by an educator that resulted in what was called the Two Sigma problem. Basically, the Two Sigma problem states that individuals perform better with one-on-one tutoring than they do in a group session. And essentially, to date we haven't had a scalable way of addressing the Two Sigma problem, but things like Atlas or AI coaching in general really get to the heart of the Two Sigma problem, and we're seeing that with things like 97% helpfulness rating or the impact that Atlas is having on someone's learning journey, which eventually leads to, potentially, a promotion or a scope increase in the job or a salary increase, and then also, more importantly, a real ROI created for the business.

McCarty Carino: So as you noted, you do also offer human coaching. So how do you think about kind of where the human touch is needed, where the AI can take over, how you leverage the strengths of each?

Singh: We built Atlas with three concepts in mind. First, it's built around a large language model. But the first concept is, we built it to be Socratic, so it doesn't give you an answer. Unlike a ChatGPT or Claude, it doesn't just give you the answer. It guides learners actually to the solution rather than just giving them the answers outright. The second one is, Atlas is built around context, personal context, so Atlas knows what the employees are trying to learn, what they're currently learning, what they've learned in the past and what they have yet to learn. And then the third principle that Atlas is built around, and this is captured both in the experience as well as from a model standpoint, it's built around being able to hand off to a human seamlessly. And so where we see humans really, like, leaning in terms of Atlas is things like motivational support. It might be things like where you need a human perspective to define a more technical concept, where Atlas is just sort of giving them more and more content, but that's not enough. Like, you need a human touch to explain something.

McCarty Carino: A big part of Multiverse's mission is to increase access to opportunities, often in fields that required a college degree and were kind of off-limits to a big part of the population of workers. How does this AI tool help to further that mission?

Singh: One thing that we see higher demand for an Atlas and the tool is among the over 40s, like employees and learners that are over 40, that may have either never gotten a traditional college degree or have been out of a college environment, a learning environment, for a large portion of their life because they've been working. The other area that we see a large usage of Atlas is with learners with additional learning needs. So those are sort of two very specific examples of where we see a tool like Atlas making our platform more equitable.

McCarty Carino: There's been a lot of talk recently about kind of diminishing opportunities for young workers, including even those with college degrees, just sort of due to a lot of changes that have taken place in the workplace. Some demographic related, some because of remote work — but in some cases, because AI is being implemented in ways to take over the types of routine tasks that often entry-level workers in a field would do. Is AI kind of a double-edged sword for this mission?

Singh: I don't believe that AI alone, at least today or in the near future, is taking jobs from people. I think it's people who have AI skills are taking jobs from people who don't have AI skills. It sounds like a pithy statement and sort of an obvious statement, but I actually think it reinforces why Multiverse exists, which is to allow young people to actually build up their AI skills, along with whatever skills they're bringing to the table already. And that is what enables them to be more successful, but also grow in the workforce. So I actually think tools like this are absolutely necessary to upskill everyone on AI because it's the people who aren't upskilled on AI [who] are the ones at risk of not having an opportunity in the new world that we live in. But people with AI skills, I think, will continue to have similar opportunities as they've had in the past.

McCarty Carino: So tell me more about where this might be going in the future. How might workers kind of continue to team up with AI?

Singh: What we're finding is, in the last couple of years, the shelf life of technical skills has dropped to about 2½ years. Imagine an AI coach that actually knows your job and is proactively pushing you to learn new things, like pushing either new content or connecting you with new people. And so I think that a tool like Atlas is something that becomes even more critical in this world where the half-life of technical skills is shrinking dramatically. I think that's the biggest example I could probably think of why AI is supernecessary in this world. I think another statement we use at Multiverse is, “AI is going to solve some of the biggest problems created by AI.” So it sounds like an "Inception" moment, but I truly believe that that's actually going to happen.