How much more content does the world need? With enough time and the right subscriptions we can access almost any content we could possibly want or need. Want to learn a new word every day? Absorb the leadership musings of a Fortune 500 CEO? Take a deep dive into an obscure field of computer science? As Apple said 15 years ago, there’s an app for that. Knowledge is just a short search and a click away.

But there are problems with this smorgasbord of easily available knowledge. It can breed short-term thinking. After all, if I can find anything I want whenever I want, why learn anything new? I can take what I need — a snippet from here, a diagram from there — apply it to the job at hand and then consign it to the recycle bin in my brain. And while this approach may accomplish what I need right now, it encourages compartmentalization.

What are we really learning when we take this approach? How to search effectively? How to mix and match? It’s quick and easy to acquire information, but learning is about more than that. It’s about striving to apply what we’ve learned and developing the skills and behaviors that help us apply that knowledge more effectively and impactfully over time.

But reinforcing learning can be a complex task, requiring numerous touchpoints and the ability to assess progress over time. And few of us have the resources to pair employees with a dedicated learning mentor, let alone at scale.

Let’s look at how artificial intelligence (AI), specifically agentic AI, is changing the game.

Mentoring at Scale

AI can assist in the development of complex, granular learning pathways in a fraction of the time it would take a human, and it can be used to scale that capability to deliver learning to hundreds or thousands of individuals. To close the loop of the learning and development (L&D) experience and drive real behavior change, you also need to stay close to those individual learners, assess varying levels of competence and deliver regular feedback. AI can help with this piece, as well.

An AI agent with the right brief is able to adapt to skills gaps, challenge responses, role-play scenarios and give feedback in the moment. It can do what any great teacher does: bring together the content with the information you need, the context to help you understand how it’s relevant to you, and the conversation by which you shake out the details, exchange viewpoints and ultimately find the perspective to move forward.

Content isn’t the bottleneck here. Content abounds, but great teaching is hard to find. So, the first hurdle a teaching AI agent needs to overcome is credibility. We’ve all spotted the occasional discrepancy in an AI tool’s results. They make mistakes. They don’t always have a discerning eye when it comes to choosing their references and supporting data. But the beauty of a digital system like this is that we can limit its scope to ensure a high quality of supporting data. We can design training environments that use only approved data from within the business or from certain approved and trusted suppliers of information. This is not a large language model, sucking in content from countless billions of sources; it can be tailored to be as discerning as we need it to be.

A Model Tailored Toward Skills Growth

Agentic AI helps to create an environment that prioritizes skills growth by identifying incremental improvements in the skills that support business success. We’re long past the point of anyone thinking that learning for learning’s sake is useful, especially when it’s so important for L&D teams to prove return on investment. So, tools like this become doubly valuable: they act as mentors to every individual in the organization and, at the same time, they focus on the skills that the company needs to foster in order to thrive.

And then there’s the human component, which acts as both oversight on how the agent is deployed (the aforementioned ability to ringfence which data the agent works with) and how successfully it’s delivering on its objectives. AI agents can be tasked with creating an ongoing, skills-based program of learning, tailored to a clearly articulated strategy of what the business requires to grow. But the final proof of that program’s success still has to be the people who see it in action. Line managers and senior leaders will be the arbiters of that success: in the specific skills targeted for growth and in the rate at which individuals grow. A salesperson can role-play a dozen conversations with an AI agent to hone their ability to close a deal or overcome specific objections, but the value of that will only be seen in actual meetings with potential customers.

Prompts and Motivations

Dr B.J. Fogg identifies three pillars for effective behavior change: prompts, motivation and ability. The right tools usually supply the prompts. In this case, the AI agent can and will learn dynamically from the learner — how they prefer to be spoken to, when they’re most likely to be receptive to learning — and prompt them accordingly. They’re able to follow up and reinforce, ensuring that they are retaining what they’ve learned.

But they can also supply a large piece of the motivation, for example by showing how growth in skills can be applied in appropriate scenarios. This is the antithesis of learning for learning’s sake in the sense that everything can be tied directly to a real-world situation. For instance, there’s a role-play for selling a new product to a potential customer in the public sector. Or there’s an exercise in discussing a second consecutive poor performance review with an employee.

As for the ability: that’s precisely what agentic AI is changing. We’re beginning to see that we don’t need to give people more and more learning modules to complete. We can actually teach them, in a relevant and measured way, in the flow of work and at scale. It’s an exciting time.