Despite its capabilities, the Learning Management System (LMS) has some inherent limitations. 

It’s an efficient delivery method for discrete learning experiences that are completed in the span of minutes, hours, or days. However, we know that it’s unrealistic to believe that any human can complete, for example, a 40-minute eLearning module on sales techniques, internalize the content, and immediately apply the new techniques with customers. After all, we forget 50% of what we learned 20 minutes after a learning experience, and over 75% after a month.    

The 70-20-10 ratio—or what we learn through doing, learn informally from colleagues and managers, and learn through formal training, respectively—isn’t a hard-and-fast measure. But using it as a starting point, we can estimate that our LMS is designed to deliver only about 10% of the learning our people will do on the job. That means it can’t account for about 90% of their on-the-job learning.

Built-in generative AI (genAI) offers a potential way to impact that 90%. For example, within a spreadsheet program, a digital assistant might pop up and offer to set up a table or equation. In this scenario, the learner never needs to interrupt their flow of work to search the LMS for a course that shows them how. Thanks to its ability to interpret a question and provide an answer (what we call conversational learning), genAI is well positioned to respond to these flow-of-work learning needs. 

Staying close to the “action” of the learner’s flow of work and using their behavior as context is critical. As learning needs intensify in quantity and complexity, we must increasingly think outside the LMS for fresh opportunities to deliver moment-of-need learning and skilling to our people. Custom AI tools, embedded within the flow of work and trained to interact with learners in a variety of ways, are an efficient and effective answer to this need. 

What’s Ahead for the LMS

In the past few years, LMS providers have been adding AI to their platforms, with emerging capabilities falling into several key areas: 

Content generation: This capacity allows L&D teams to instantly create eLearning modules from information in uploaded resources and a series of prompts, often with click-to-publish functionality. For organizations that view the L&D function primarily as a distributor of training, this feature may be appealing—it eliminates the weeks or months required for typical learning content development. Caveat emptor: The AI-generated content is often generic and lacks the depth required for true skill development.

Labeling and indexing: AI’s ability to generate course descriptions, titles, and tags helps learners to locate and access the content they need within our vast (and growing) libraries of learning content. These tags can often be cross-linked to skills taxonomies used by managers, HR, and recruiting teams: a feature many organizations are embracing.

‍Administrative assistance: Modern LMS platforms are increasingly incorporating agentic AI that can take real action within the world. Instead of navigating complex menus on the back end, an LMS administrator might simply type, "Enroll all learners with these attributes," and the AI agent adds the resource to the learning paths of all learners who meet the criteria. 

Agentic AI is another feature currently “under construction.” Nonetheless, it may evolve to the point that LMS administrators can simply describe a desired future state, and the system handles the underlying configuration to make it happen. 

Matching resources with needs: L&D leaders have long been seeking a Rosetta Stone that will evaluate learners’ skills and needs and supply them with existing resources that address their needs and gaps. Performance LMSs (PLMSs) can embed skills frameworks that use AI to draft competency models based on job descriptions, then use this model to push the right content to learners who need it. 

Knowledge tracing: Instead of starting with the simplest concepts—and forcing experienced learners through introductory content—knowledge tracing gathers information about each learner as they proceed through training, observing their performance, factoring in individual differences such as role and experience, and providing remediation when they encounter difficulty. 

In addition to tailoring learning experiences to their specific needs, knowledge tracing offers learners agency and autonomy. They learn according to their needs and preferences, rather than being forced to work through a one-size-fits-all learning journey.

It’s an ideal way to learn, but it’s difficult to do well: Stay tuned for the new opportunities technology may offer in the years to come. 

Adaptive learning: In our pre-AI world, it was a challenge to achieve truly adaptive learning that responded to learner’s skills, understandings, and needs. L&D leaders have often been constrained by budget, need, and timelines to a more one-size-fits-all approach. However, emerging genAI functionality in some LMSs can analyze how learners answer assessment questions, as well as information about their role, to tailor a curriculum to each learner. 

This functionality is a promising move! However, the reality hasn’t quite caught up with the promise. Current implementations frequently rely on superficial data points like job roles or interests rather than deep skill analysis. Still, innovations that promise greater flexibility and truly personalized learning pathways may not be too far in the future.

Where Does the L&D Leader Fit into the Picture?

LMS capabilities will continue to evolve and, ideally, become easier and more intuitive to use. It’s not necessary for L&D leaders to develop a deep understanding of their architecture or engineering, but we should be up to date on emerging LMS features, their strengths and weaknesses, and how LMS functionality affects our efforts to upskill our people. 

‍In 2026, let’s move toward ungating our learning and delivering more in the flow-of-work, making it freely available within the platforms and workspaces our people use. 

From Content Creator to Strategic Advisor and Curator 

As we shift from a content generation-focused role to one more focused on AI and governance, we move away from traditional learning design and development, which yields static solutions, toward a more iterative process, which yields dynamic, adaptive learning solutions. 

These newly dynamic and adaptive solutions are founded in a knowledge base composed of our organization's values, principles, standard operating procedures, terminology, and so on. The creation of this knowledge base, a process called context engineering, combines Retrieval-Augmented Generation (RAG) and prompts to ensure that AI output is relevant, accurate, and tailored to learners. (Return to Trend 5: The L&D Leader in the Technology Ecosystem for more on this process.)

Amid the cacophony of AI-powered bells and whistles, it can be a challenge to sort out the functionalities and features that matter most. Reach out to share your learning tech goals and challenges…and get a tech-agnostic take from our experts.