As organizations look for more personalized, intelligent and connected learning experiences, artificial intelligence (AI) is changing what learning management systems (LMSs) can offer. AI-native platforms are driving new ways for how employees access learning and how L&D teams support workforce development.
In this episode of The Business of Learning, sponsored by Harbinger Group, Poonam Jaypuriya, vice president of eLearning at Harbinger Group, discusses how LMS platforms have evolved, what it means for a platform to be AI-native and how L&D leaders can prepare for the future of learning.
Check out the episode now to learn:
- What sets an AI-native LMS apart from traditional learning platforms
- How AI can personalize learning based on employees’ roles, skills and business needs
- What L&D leaders should prioritize when selecting a future-ready LMS
More Resources:
- [Toolkit]: Managing Your Learning Tech Stack
The transcript for this episode follows:
[Ad]
Learning is changing fast, and so are the platforms that support it. As learning becomes more connected, contextual and intelligent, your LMS needs to evolve too. At Harbinger Group, we bring over 35 years of experience in learning and technology to help you make the right strategy and technology decisions. From digital platforms and LMS integrations to AI-powered learning and custom e-learning content, we build connected learning ecosystems that strengthen capability and drive workplace performance. Ready to evolve your learning ecosystem? Visit www.harbingergroup.com.
Michelle Eggleston-Schwartz: Hi, welcome back to The Business of Learning. I’m Michelle Eggleston Schwartz, editor-in-chief here at Training Industry.
Sarah Gallo: And I’m Sarah Gallo, a senior editor. The learning management system has changed significantly over the years, and now AI is bringing new possibilities for what these platforms can do. So today we’re exploring how LMS platforms are evolving and what these changes really mean for learning and development leaders. We’re joined by Poonam Jaypuria, vice president of eLearning at Harbinger Group. Poonam, thanks for speaking with us today.
Poonam Jaypuria: Thank you so much for having me on this podcast today.
Michelle Eggleston-Schwartz: Yes, welcome, Poonam. Excited to have you here. To get things started today, I’d love to hear, in your opinion, how has the LMS evolved over the past five to 10 years? What do you think has driven the biggest changes in organizations’ expectations around these learning platforms?
Poonam Jaypuria: Yeah, that’s a great start, Michelle. Let’s look 10 years back. The LMS had a fairly well-understood job. Everyone understood the LMS as a platform that would host courses, enroll learners, track completions and certifications, and handle some compliance. The L&D team would get reports from the LMS. That’s really what the LMS was. Those things haven’t gone away I would say. Even today, they all still matter. But I think expectations around the LMS have changed dramatically over the last couple of years. The first change I see is simply how work is happening now. Employees don’t spend their day inside an LMS. They’re mostly on Teams, Slack, Salesforce, ServiceNow, HR systems, development platforms and AI copilots now. This idea that I’ll stop work, go somewhere else, search for a course, take it, and then come back and continue working in a disconnected learning environment. I think people are not working that way anymore. That’s one big change that is happening. The second change is the pace at which knowledge and skills are changing. In fact, there was a World Economic Forum report that came out recently saying that almost 39% of workers’ core skills will change by 2030. That’s significant change and skills disruption that we’ll all experience over the next few years. What’s happening is that the shelf life of knowledge is collapsing. I remember a time when you would build a course and expect that, for the next two or three years, there would hardly be any updates, and people would take the same course. But today, the speed of development is so fast that literally every other week, you hear about something new coming out. A product feature we were using today might change next week. Regulations are changing at an equally fast speed. So if you documented something in January, most probably it was outdated by June. That speed of change in skills and content is a big disruption we’re going through. The third change, which I think is probably the most important, is that the organizations have started asking a bigger, or better, question of learning teams. It’s not so much, “Did Poonam complete the course?” The question is more, “Can Poonam really start doing the job, or can she become productive at work faster?” I think that distinction of completion and readiness is the biggest change I see, even personally here at Harbinger. I would argue that, with this, the role of the learning management system has to evolve from completion tracking to how do I make it give insights on readiness. That’s the question the LMS platform has to respond to. I’ll give you an example. I recently met with a lot of professional education companies, associations and workforce learning organizations in the U.S. What came out is that very few of my conversations were, “Oh, we need more courses. How can we scale up development?” None of that. People were asking, “How do we help someone actually practice what they’re learning?” or “How do we know that someone is ready for the job?” or “How do we know that, if AI is being used, it’s drawing from a trusted knowledge base?” How do we make all the content and expertise they have sitting in webinars, courses and documents much more useful using new technologies? I think that’s the evolution of the LMS I would say: from learning administration to learning experiences, and how increasingly toward performance and capability enablement. So I wouldn’t describe the biggest evolution of the LMS as a list of features. I would say the purpose of LMS is changing from managing learning activity to performance enablement and workforce capability. Once you make that shift, you start expecting a very different kind of learning platform.
Sarah Gallo: Great points, Poonam, and something we’ve seen in our coverage here at Training Industry, too. People aren’t just wanting to add more courses and completions. They want to know that what they’re doing is actually working and driving performance. So I love that point. I think another evolution we’re hearing a lot about right now is around this AI-native LMS platform. Could you share what that means and how it’s different from a more traditional LMS?
Poonam Jaypuria: Yeah. Interestingly, the term “AI-native” is being used loosely in many ways. I want to be very careful, because at literally every conference I’ve attended in the last two years, everyone is calling their platform an AI LMS or an AI-native LMS now. What happens is, when you add a chatbot, use AI to generate questions, summarize a course, generate a course or AI-powered search, those are all great AI features, but I wouldn’t call them an AI-native learning platform. Those are more AI features. For me, a simple way to test whether something is AI-native or AI-enabled features is to ask: Is AI a feature inside the platform, or is it an intelligence sitting on top of something that’s already there? If you see that distinction very clearly, it will help you understand what’s AI-native. Let me give you an example of a large healthcare learning organization we’re working with. They have a very mature learning platform and a complete ecosystem. It’s not just an independent platform. It integrates with huge systems and stores a lot of data. A few years ago, many of our conversations with them were around platform engineering, modernization of the platform, automation and content operations. All of that is still important. But today, the conversation is very different. We’re discussing things like: How would learning and performance data influence my next learning experience? How can trusted content become available to AI? How do you understand enough about the learner and their situation to provide useful support through the learning platform? How do you connect learning, skills and business context together to deliver not just better, but more relevant learning experiences? And how do you govern all of this to ensure responsible AI and you make sure that there’s no hallucinations? What’s interesting is that none of these questions starts with, “Where can I put an AI feature?” They start at the architecture of the platform. For me, that is the real shift of AI-native. I often think of the architecture as a simple flow. There are five key things you should worry about: trusted knowledge, context, intelligence, action and the experience. First, the system needs trusted knowledge, which is where content adds a lot of value. What does the organization actually have? What are the current documents? What’s approved? What is outdated knowledge? That’s trusted knowledge. Then comes context. Who is this person? What are they trying to accomplish? What do they already know, so I don’t have to train them again? What situation are they in currently, and what do they want to achieve? That’s the context about the individual. Then comes intelligence. Given the knowledge and context, what should happen next? That’s the intelligence layer. Then comes action. Should I recommend something? Should I generate practice assignments? Should I create a learning pathway? Should that person be connected to an expert, or should I even alert the manager that this person is ready for the next role? That’s the action the system should take. And finally the experience, which is what the employee or learner actually sees. The way I look at it is, my favorite question when someone tells me their platform is AI-native is: If your approved knowledge changes today, will tomorrow’s AI experience know that the old information is no longer valid? If the system answer this, then you can say it’s an AI-native platform. If someone says, “Oh, I can get this information from a quick search or GPT,” that’s not AI-native for me. I don’t think AI-native is a binary product distinction. You don’t become AI-native just because you have some AI chatbot in your LMS. For many organizations, the right answer may actually be to retain the LMS they have today for certain functions and modernize the knowledge, data and intelligence around it to create an AI-native experience. I’d define it as a maturity journey rather than more like asking, “Can I buy an AI-native LMS?
Michelle Eggleston-Schwartz: Definitely. Thank you for breaking that down, because there are key differences. It’s not just features. It’s something much more robust. It’s the architecture, as you mentioned there. I’d like to touch on the learner experience and the learner perspective here. From a learner’s perspective, what will an AI-native LMS do differently? What are some of those AI-powered experiences or capabilities employees can expect to see in this AI-native platform?
Poonam Jaypuria: For me, as a learner myself, one of my success criteria for an AI-native experience is whether I can spend less time finding what I’m looking for and what is relevant to me right now. That’s my best way to describe an AI-native learning experience. Today, all of us as learners carry a lot of expectations. First, we have to solve a lot of problems. I have to decide what I need to learn, go to the LMS, find the courses and spend 45 minutes or an hour taking a course. In this process of learning, AI has changed what’s possible almost overnight. What if the future of personalization isn’t just better recommendations, but better practice? This became very real for me during a recent conversation with a professional association. Their members work in situations where they have to make decisions using judgment. The answer isn’t always usually just right or wrong, or black and white. When we started discussing an AI-enabled learning experience, the most interesting idea wasn’t “How can we recommend the right course,” necessarily. It was, “Could we give professionals a safe environment to practice difficult situations before they encounter them in the real world?” Give them a realistic scenario. Let them decide what they would do. Allow them to ask questions. Let the scenario change based on their response and then eventually provide feedback grounded in the association’s trusted knowledge and guidance. That’s where AI starts becoming a differentiator. It’s not just about just delivering more content or better recommendations, but also creating opportunities to think, decide, practice, get feedback and try again. We’re exploring similar ideas with other L&D teams and education organizations, and it has really changed one part of my thinking. For years, we’ve talked about personalized learning, and I think AI is definitely making that much easier. But increasingly, I think the more valuable question may be: Can we personalize capability development rather than just personalize content? Maybe I don’t need another course personalized for me. Maybe I just need 10 minutes of practice. Maybe I need a coach. Maybe I just need a job aid, and I’m good to go with my next steps. There’s another related shift happening specifically in certification, specifically. The question is changing from, “Did this candidate complete the exam prep content?” to “Is this person actually ready to take that exam and hopefully clear it?” Once you ask that question, your learning architecture changes. You start thinking about better diagnostic assessments, targeted remediation, practice, confidence building and simulations, rather than just saying, “Here’s a personalized learning pathway for you.” From that perspective, I think learners are moving from searching to asking questions, from looking through course catalogs to conversations, from fixed pathways to adaptive support and perhaps from content consumption to practice and readiness. In fact, LinkedIn recently published workplace learning research that found that 91% of L&D professionals believe continuous learning is more important than ever for career success. I think AI gives us the opportunity to make continuous learning feel more like a part of my day rather than a series of separate learning events.
Sarah Gallo: Definitely. That opportunity for not only practice, but realistic practice, especially for those human skills like judgment that you mentioned, is really huge. I’d love to dig a little deeper into the personalization piece here. As LMS platforms become more intelligent, how important do you think it will be for them to understand a learner’s broader context, such as their role, skills, current projects and business needs? What does that look like in practice?
Poonam Jaypuria: I think context may become one of the most important capabilities of the future learning ecosystem, because AI or intelligence without context isn’t really personalization. Let’s take Netflix as an example, which we’re all familiar with. If I’ve watched three thrillers, it can recommend another thriller. That’s great. But in workplace learning, I think it’s more complicated than that. Knowing that I’ve completed a leadership course doesn’t tell me that I can manage 100 people. It doesn’t tell me whether I’m a good manager or not. Completing a course just doesn’t tell me enough about whether I’m ready to take on that role or opportunity. Let me give you an example. Let’s say two employees both have a gap in presentation skills. One is presenting an internal product update next week, and the other is pitching a multimillion-dollar deal to a customer. Technically, they have identical skill gaps, but in reality, the intervention will look very different for both. That’s the context we’re talking about. It’s not just who the learner is, but what they’re trying to accomplish, under what conditions and using which trusted knowledge. That’s a richer definition of context for me. This is where learning may have to connect with the broader enterprise ecosystem, because some of my context will be in an HR system: my role, tenure, skills and career path. Some may be in my CRM: the customers I’m talking to, the opportunities I’m working on and the products I’m selling. Some may be in my project system, some in my performance management system and some in the LMS. My work spans multiple systems. We’ve already seen a market signal here. A few weeks ago, Workday made a big announcement that Workday Learning would now be powered by Sana, which was generally available. What I find significant isn’t simply that there’s now AI in Workday. It’s that the learning experience will sit on top of Workday’s trusted database of people’s skills and other information, so learning can be grounded in workforce context. That’s the direction I think the whole industry is heading. The LMS doesn’t need to own every piece of context, but it needs to use context from across the organization intelligently and, of course, responsibly. That raises a harder issue: permissions. Just because data exists doesn’t mean the learning system should have access to all of it. Organizations really need to think about what decisions they’re trying to make with learning context and what data AI should have access to. What should employees see? Where does the human come into the loop? Those are the questions that should shape the design. If we don’t think about those questions carefully, personalization can very quickly start feeling like surveillance. I think one of the disciplines L&D leaders will increasingly need to understand is not just learning design or AI, but context design.
Michelle Eggleston-Schwartz: Such a good point. Context is everything, and as you said, if it doesn’t have that context, it’s not true personalization. Great points there. I’d love to hear, for learning leaders out there who are considering a new LMS, what advice would you give them? What capabilities or questions should they consider to ensure they’re making the right purchasing decision?
Poonam Jaypuria: It may sound a little counterintuitive, but my first piece of advice would actually be: Don’t begin with the LMS. Begin with the business problem. It’s very easy to start an LMS evaluation by rolling out an RFP with a 150-feature list. Most of us have heard about a lot of AI features by now, like AI search, AI authoring and AI-generated quizzes. It’s easy to put that checklist together, and those questions may matter, but maybe not to start with. I’d start somewhere else. What are we actually trying to achieve to help people do better? Is it faster onboarding, sales readiness, compliance, certification readiness, manager effectiveness, workforce skilling or reskilling? Two companies can buy the same LMS and get completely different value from it. Then I would ask vendors some hard questions, like: How does your AI know what to trust? What is it grounded in? Can I define the authoritative sources? Can I distinguish between approved and unapproved content? Can I trace an answer back to its source? Those questions have become important in conversations we’ve had with our customers, because almost every organization owns a lot of intellectual property, but often the expertise is locked inside documents, courses, videos and subject matter experts’ knowledge. If you want your AI to use that knowledge intelligently, you need to think beyond courses. You need to start thinking about content as knowledge infrastructure. What’s the structure of my content? Can I break it into useful components? Can it be tagged? Can it be versioned? Can the organization determine what AI should and shouldn’t use? Those kinds of questions help you make a better choice about which LMS will work for you. The second set of checks I’d recommend is: What context can the platform understand? Can it go beyond my learning history? Like the example we discussed, because I took two leadership courses, will it just recommend another one, or is there more to it than that? Can it use my role, skills, context and workflow signals to actually help me find the right content? Can it integrate with my other systems, where that context already lives? As we said earlier, learners aren’t going to go to an LMS to learn. Can learning come to me where I am, where I’m spending my workday? That’s my second set of guidelines for L&D admins. The third is: Can AI just recommend, or can it also act? That’s a big difference. A lot of platforms can say, “You may like this course. It’s recommended by AI,” it can create a pathway for you. But as systems move toward action, permissions and controls become important. Can the platform surface important information and create a workflow for me based on my preferences? That’s where I’d ask whether AI can recommend or act as well. That’s the third thing I would look for. The fourth is: How is AI evaluated and governed? That’s a big one. We’re all hearing stories about how AI can make mistakes. I would actually ask my vendors: How do you know an answer is good? What happens if the AI doesn’t have a good confident answer? Can I see the source of the content from which AI generated the answer? How does human approval come into the picture? What is AI explicitly not allowed to do? We’ve been working on AI concepts in professional and high-stakes learning environments recently, and this was very important. Sometimes the best AI answer is, “I don’t have an answer.” Sometimes it’s, “Here are the sources, but here are the issues to consider. Can you get someone involved to make the right decision?” If AI can do that, I’d say it’s good, trusted AI. The fifth and last one is: How open is the architecture? Maybe because I come from a technology background, this is huge for me. I wouldn’t try to predict which AI model will dominate three years from now. We all know how things have evolved in just the last six to eight months. I don’t think anyone knows which models will be there three years down the line. Can I change models? Can I connect new knowledge sources? Can I bring in external data? Can agents securely interact with my platform? How open is the architecture from a long-term perspective? I’ll give you an example. I was talking to another customer who had a major AI coach using a certain model just eight months earlier. They got a lot of good usage out of it, and suddenly it was an old model. They had to completely retrain on a new model, putting in another three months of effort. That’s not scalable. That’s why understanding openness in the architecture of the platform will be key from a long-term perspective. In other words, don’t future-proof by predicting the future. Future-proof at the architecture or design level. And I’d also ask about outcomes. Producing more content is useful, but more content isn’t necessarily the business problem I’m trying to solve. I’d rather focus on performance and readiness and ask questions along those lines.
Sarah Gallo: Definitely. Starting with the business challenge you’re looking to solve is huge, so thanks for sharing those questions. You mentioned earlier how today’s employees are working across different systems, platforms and tools, sometimes all within the same day. I’d love to know, from your perspective, in the future, do you still expect the LMS to serve as the central hub for learning, or do you think learning might happen across a more connected ecosystem of tools?
Poonam Jaypuria: Yeah, interesting question. Earlier this year, while we were strategizing at Harbinger, one thing I said was that the LMS is eventually becoming infrastructure rather than a platform for me to go to and learn. I know that’s counterintuitive, but that’s where I see the LMS heading. I don’t think LMSs will disappear because we still need learning records, certification, entitlements and reporting. All of that is critical for us and probably will remain even more critical. But the assumption that learning has to happen inside the LMS is weakening. A recent customer conversation brought this point out. We were speaking with a prospective customer, the president of a company whose people have deep technical expertise and years of customer-facing experience. One of the opportunities we discussed was: How do you make that organizational experience available to someone preparing for a customer conversation? Rather than thinking, “I should go to the LMS and take some training,” the experience should be built around the work the learner is performing. Before a customer meeting, the system could surface relevant knowledge, previous insights about that customer and questions worth considering so I’m better prepared for the conversation. After my customer meeting, it could ask: What did I learn? What was something new that surprised me? Was there something I couldn’t answer? Could some of that insight become knowledge for somebody else, so the next person coming into the system knows how to prepare better for those customer meetings? Think about what happened there. There may still be a learning platform sitting somewhere, but from the employee’s perspective, it doesn’t feel like, “Oh, I went to an LMS and I learned.” The Learning became part of the work. I think we’re moving much more toward that. The LMS will connect with knowledge systems, HR platforms, CRMs, skills systems and AI assistants. The intelligence doesn’t necessarily have to be inside one LMS or one system, and perhaps the learner doesn’t even need to know which system provided what piece of information. Let me give you a banking analogy. Twenty years ago, if you wanted most banking services, you would go to a bank branch. Today, banking happens everywhere. You transact through your phone or inside another application. Maybe financial services are embedded in other experiences. But the bank didn’t disappear. It became the infrastructure behind the experience and, in fact, it came closer to us than it was before. I think something similar will happen with learning. Which is why I don’t think the question is, “What will replace the LMS?” The question is probably, “What role should the LMS play when learning no longer happens at one destination?” Earlier, I might have described the LMS as a system of record. Then it became a system of engagement, and now it’s becoming a system of intelligence, moving toward connected capability infrastructure. The intelligence itself may increasingly sit outside the LMS ecosystem. That doesn’t make the LMS less important; it just changes where its value sits.
Michelle Eggleston-Schwartz: Well, before we wrap up today, I’d love to hear one final piece of advice you would give learning leaders who want to make sure their LMS is future-ready.
Poonam Jaypuria: I would say don’t make the LMS the centerpiece of your learning strategy. Make the learner and the business problem the center. Don’t start by asking, “How can AI help me improve my LMS?” Start by asking, “Where do people need better judgment? Where do they need more practice? Where are they struggling? What do they need help with while actually doing the work? And where do we genuinely need a course?” I think that last question is important, too, because not every performance problem needs learning content. Once you understand those moments, you can then decide what the LMS should do, what an AI assistant should do, what a knowledge platform should do, what an AI bot should do or even whether humans will add value to the learning experience. The technology is going to keep changing. AI will keep changing, models will change and interfaces will change. But what will give an organization a good foundation is trusted and usable knowledge, connected data, clear context and an open architecture that can be future-proofed. If those foundations are strong, I think you’ve made good choices. Traditional learning usually asked, “What courses should this employee complete?” I’d say the next generation should ask, “What capability does this person need right now, and what’s the best way to help them build that capability?” That’s the future of learning platforms, and that’s what I’m most excited about.
Michelle Eggleston-Schwartz: Awesome. Well, thank you so much, Poonam, for sitting down with us today. How can our listeners get in touch with you if they’d like to reach out?
Poonam Jaypuria: Absolutely. You can easily find me on LinkedIn. I’m Poonam Jaypuria. You can also reach me through the Harbinger Group website. And I’d genuinely love to hear from learning leaders who are wrestling with these questions. We’re in very interesting times right now, and technology is moving so quickly that organizations are still trying to find out what good learning will really look like. Some of the best ideas, I believe, come from sharing what we’re learning from these real-world experiments.
Michelle Eggleston-Schwartz: For more resources on this topic, check out the episode description or visit the show notes on our website at trainingindustry.com/podcast. And don’t forget to rate and review us wherever you tune in to The Business of Learning. Until next time!
