AI in Learning and Development (2026)

L&D has a measurement problem. It's hard to prove the connection between investments in learning and business outcomes (or at least hard to make that case neatly to your finance team).

In a 2025 survey, nearly 1 in 3 leaders admitted they couldn't connect their L&D investments to their company's revenue or profit margin.

That's a problem, and AI is only exacerbating it.

Take manager training, something I’ve run dozens of times. Years ago, it was easy to get leadership buy-in for manager development. Not only was everyone doing it, but there was an entire body of research demonstrating that managers drive performance and retention.

Today, it's much harder to make the case to invest in middle management, especially when you have companies trying to flatten their hierarchies, or replacing managers entirely with AI. But it’s also an opportunity to reconsider if your management training was really working.

If that’s causing you an existential crisis about your job description, you’re not alone. AI is raising the stakes for how we measure impact and prove our value to the business. I think it’s also giving us an opportunity to rethink what value we bring as learning experts.

How is AI used in training and development? 

87% of L&D professionals are already using AI according to our 2026 report. How they're using it, however, varies greatly. Some organizations are still experimenting, while a handful are systematically integrating it into their workflows.

Here are four trends I'm seeing in how L&D teams are using AI.

1. Speeding up content production

84% of L&D teams are using AI to speed up content production. That includes tasks like drafting content and quizzes (60%), generating narration (63%), creating videos (52%), and localizing materials (38%). Meanwhile, only 39% report using AI to strengthen their strategy.

Note: our survey had 421 respondents, many of whom are either our customers or in our networks. This likely skews the AI adoption rates higher.

2. Implementing adaptive learning paths 

33% of L&D teams are planning to implement personalized learning pathways in the next 12 to 18 months. The good news is that you don’t need to necessarily design any new content to get started. 

That’s because you likely have eLearning content sitting somewhere, whether that’s in an underutilized LMS/LXP or even unpublished in an authoring tool. It might be structured as standalone courses or fixed programs with pre-determined milestones, like assessments. 

With tools like an AI-native or headless learning platform, those eLearning components can be transformed into dynamic experiences based on the user's role or prior experience. That means you no longer have to choose between scalability and personalization when designing learning experiences.

3. Delivering training in the flow of work

Josh Bersin coined "learning in the flow of work" in 2018 to describe employees getting the information they need, when they need it, without switching contexts. For that to work at scale, training has to be modular. 40% of teams are already using AI search and knowledge assistants to deliver on this.

With AI, we're closer than ever to closing that gap. Consider a nurse preparing to perform a procedure they haven't done recently. Before entering the room, they can search in natural language, "how do I do this procedure," and get the right guidance immediately from their organization's vetted knowledge base.

Assets are deployed at the moment of need, with flexible delivery standards so you can track what's working and adjust what isn't.

4. Predicting skill gaps

According to LinkedIn's 2026 Talent Velocity Advantage Report, 86% of companies report needing better insights into the skills their organization needs to adapt and grow.

That's not news. We're always seeking ways to better anticipate the demands of the industry and where it will go. That means we not only need to measure our workforce, but also get better at prediction.

What that requires is tooling that is excellent at pattern recognition. We're not always going to get this right, but what we can do is use AI tools to create pattern-based predictions grounded in historical data. That's what it is good at: processing massive data sets and training on them to anticipate subtle correlations, without the bias we bring.

The question is where to start. The use cases below are the ones where L&D teams are seeing the most traction.

6 ways to implement AI in your L&D strategy

It's tempting to just start experimenting with AI tools, and if you haven't already, there are plenty of examples of how your peers are already using AI below.

But before you start experimenting, I'd challenge you to approach it more strategically.

One of the most common mistakes I see is teams starting their AI implementation with tools. A shiny new tool isn’t going to solve all of your problems, especially if you’re already struggling to demonstrate how your work is connected to the business. 

Instead go back to your strategy and roadmap. Find one learning experience that is in need of a refresh or complete overhaul. Using your evaluation data, determine what needs to change, and why. Then document where AI will show up in the workflow, who will own the quality of the AI output, and how you'll know if it's improving performance. 

Here are examples of what I mean by that.

1. Content generation

It's performance review season. After being repeatedly assured “nothing” was changing, you find out the rating scale has been rewritten ever so slightly. Instead of panicking about all the places you put that rating scale, you go to your company-approved LLM, and ask it to surface any assets with the previous scale. Once you’ve located all the assets (facilitator guides, one-pagers, and the like), you use the LLM to generate updated versions. Late-night content creation session averted.

2. Localization

You’ve been asked to train technical ops teams across Latin America. That means turning regulatory content into engaging learning, and doing so in English, Spanish, and Portuguese to start. With an AI video platform, you can produce and localize training across all three languages, have a local team member verify the quality, and when the regulation changes next year, update the source once and republish across all variants.

That's how LATAM Airlines' support training team approached training their distributed employee population of 37,000, cutting production time by 83% and reaching 16,000+ learners across three languages.

3. Scenario-based role play

A manager reaches out to you because call logs show support agents struggling to de-escalate frustrated customers. They can’t pull everyone off the floor for a workshop, and you don’t have the capacity to facilitate multiple sessions. So you turn to AI for scalable role play. You feed the call logs into an LLM to generate three scenarios based on the most common escalation patterns. 

With an AI coaching tool or agent, you offer 5-minute sessions based on the scenarios. They get real-time feedback and a chance to try again, before jumping back on calls. 

4. Just-in-time support

A newly promoted manager is feeling in over their head. Their HR Business Partner is on vacation, and they haven't gone through any formal management training yet. In one hour, they have a meeting with a direct report who is going out on parental leave. They have no idea what they're supposed to do.

Fortunately, you've built a manager support agent that guides them in preparing for the meeting. They can quickly skim training resources, like a coaching guide on supporting the transition and staying connected during the leave. The agent also sends them reminders about contacts the direct report can reach out to with questions about benefits or anything else.

5. Personalized learning paths

Two new software engineers join your company. They’re starting in the same role and have the same years of experience. But one engineer has never worked in your industry before. Your AI-native learning platform adapts their onboarding experiences accordingly, offering a “new to industry” learning path to the latter engineer. 

6. Skills mapping and gap analysis

In a planning session you learn the business plans to expand into a new region in the next 18 months. The workforce planning team is looking for support mapping the existing workforce's preparedness for this transition. So you build out a skills map and use AI to cross-reference it against your current workforce data in the HRIS. From there you can identify missing capabilities, and get to work designing development pathways — before the expansion officially begins.

What are the benefits and challenges of using AI in L&D?

If you try any of the use cases I shared, you'll probably notice some anecdotal benefits, like saving your team time. But if you're looking to make the business case for embedding AI in your L&D strategy, especially if that involves acquiring new tools, you'll want to keep a few important considerations in mind.

1. Capacity

Benefits: Across the board, the most immediate gain L&D leaders report is speed. 88% of teams report saving time on content creation. Faster production means content can be delivered more closely to the moment of need. It keeps up with the business.

Challenges: Producing content more quickly doesn't necessarily translate to better outcomes. We're producing more content, but not following through to understand whether any of it is changing behavior. (I've written about this pattern as readiness debt.) The risk for L&D leaders is that AI makes it easy to look productive without being impactful.

2. Cost

Benefits: 45% of teams are already reporting cost savings from AI. That's because they can stop outsourcing specialized tasks like graphic design or video production. Those line items add up quickly, especially every time there's a content refresh.

Challenges: AI can be expensive, especially at the enterprise level. Licensing costs and usage caps can impact the way you design and use these tools (more on that in the tools section below).

3. Governance

Benefits:  Building AI governance into your workflows is an opportunity to standardize (and document) processes your team has likely been running informally for years. Things like how content gets reviewed or how updates get approved.

Challenges: Most enterprises are still learning how to govern AI. That means L&D teams can find themselves in a position where they don't quite know who owns what tool, or how to bring in new tooling.

4. Business impact

Benefits: 41% of L&D teams say AI is already contributing to business impact by enabling them to create content more quickly and meet business needs. It is also shortening the time between when we deliver content and when we can see signals of it working.

Challenges: Most teams haven't reached the maturity to demonstrate measurable business impact from AI in L&D. Only 19% are using AI for evaluation, compared to 65% using it for content development.

How do you measure AI's impact on training outcomes?

Performing a robust evaluation of training outcomes can feel like a herculean task some days, which is why sometimes we settle for NPS or completion rates. And it's likely why 63% of L&D teams need more support assessing impact.

Once you add AI into the equation, you're adding another layer of complexity. So you need a streamlined way to assess before and after AI is introduced.

To get started, you'll need four inputs:

• One observable behavior that the training was designed to change or impact
• Two signals of that behavior in the workflow: one leading indicator (something you can observe happening during or right after training) and one lagging indicator (something that shows up in performance data weeks later)
• A proxy measurement for business impact

(I know, you just rolled your eyes at signals and proxy measurement.)

Here's what I mean, using one of my use cases from above — AI coaching for customer support agents.

Let's say that last quarter you developed an eLearning focused on de-escalating calls with frustrated customers. You published it through your LMS and assigned it 1,000 global agents.

Before you can introduce AI coaching, you need to document what happened with the previous intervention. You can use a template like this:

"We expected behavior X to change. We see evidence it did (or didn't) because metric Y moved in the expected direction (or didn't). Here's what we'll do next."

In this case, that might look like:

"After completing the eLearning, we expected agents to more consistently de-escalate frustrated customers using the three-step resolution framework [observable behavior]. Managers are reporting fewer escalations [leading indicator], and the average handle time on difficult calls has decreased [lagging indicator]. Additionally, CSAT on resolved complaints improved [proxy measurement]. Nonetheless, agents are still performing inconsistently in the first 60 seconds of these calls, which suggests there's a need to focus on the first step, acknowledgment, in the resolution framework."

There's a reason that's hard to fix with a one-off course.

There's your baseline for before AI. Now you can focus on building practice and feedback loops for agents. So, you identify a select group of the 1,000 who need more practice and try out an AI coaching tool. You build a few scenarios focused specifically on the first 60 seconds of a difficult call.

After a month, you observe more consistency in opening acknowledgements in call transcripts among the pilot participants. Their escalation rate also continues to drop in comparison to the control group.

To keep costs under control (AI coaching can be expensive, y'all), you decide to make the tool available on an as-needed basis. Managers flag anyone who shows a pattern of inconsistency in acknowledgments and encourage them to self-enroll in coaching.

Finally, you track whether reps who used the AI scenarios show improvement in first-contact resolution rates over the next three months. After all, that's the business metric the leadership cares about.

If they show improvement, that's your business case for investing in AI.

What are the best AI tools for training and development?

Figuring out the best AI tools for L&D is like finding the perfect pair of jeans. You have to try on a lot of pairs to find the right fit, at the right price, that works with what you already own.

While the metaphor may be a bit simplistic, here's what I mean. Every time you add to your tech stack, you do so with a budget and organizational context. That means understanding what tools you already have (whether L&D owns them or not) and how they do or don't work together. Security and accuracy concerns, integration challenges, and legal restrictions can also slow down AI procurements.

That's why I want to start with a tool you likely already have access to, but probably don't control: the LLM.

1. Large Language Models (LLMs)

I think of LLMs as the AI workhorses for L&D teams. They're where we're able to get the most support, whether that's in time and cost savings for instructional design or in capacity freed up for more strategic work.

Most enterprise teams already have an LLM, which means it may already be connected to internal knowledge sources and (hopefully) your L&D tech stack. So whether it's Claude, ChatGPT, Perplexity, Gemini, or something else, make it work for you. 

Note: If your company is considering different LLMs, or you have the ability to select one yourself, I highly recommend reading Dr. Philippa Hardman's breakdown of the AI models for instructional designers.

2. Learning Management Systems and Learning Experience Platforms (LMS/LXP)

Whereas an LMS has traditionally been where learning is delivered, tracked, and measured, an LXP has focused more on the learning experience itself. Today, the lines between the two are blurring.

There are few AI-native LMS or LXP platforms. More often, you're seeing AI added on whether for admin tasks like enrollments and reminders, or analytics like dashboards and reporting. Where the greatest potential lies is in personalized learning experiences: the ability for employees to use the platform the way they would an LLM — asking questions, much like you would a tutor, but in a controlled environment where the information is carefully vetted.

Few platforms are able to deliver this experience, which is why I recommend Sana.

Sana supports learning in the flow of work with features like its AI tutor. Employees can search for answers in natural language, get just-in-time guidance without leaving the platform, and receive personalized content recommendations.

Works best for: Teams looking to replace a legacy system or bring in a new platform, and teams moving from static content libraries to dynamic and personalized learning.

Falls short when: Teams have a robust SCORM content library they're looking to maintain within one platform.

3. eLearning authoring tools

eLearning authoring tools are what enable you to build the content hosted in your LMS or LXP. Some platforms have an internal authoring tool, which allows you to build courses or programs with components like knowledge checks or interactive scenarios.

Many eLearning authoring tools have AI added on, and there's a structural reason for that. These tools were designed to output content in SCORM packages. SCORM is a static standard, meaning that AI cannot modify the contents. This limits your ability to adapt content to individual learners or capture data beyond completion rates and assessment scores. That's why you see platforms like Sana taking an AI-native authoring approach.

That said, one of the primary ways AI is working with eLearning authoring tools is by transforming how content gets built. Whereas before, an instructional designer would go into these tools and craft a course by selecting each component manually, now they can use natural language and existing content to get a significant head start. That's the case with Articulate 360's AI Assistant.

Articulate 360 is a suite of eLearning tools, including Rise, a browser-based authoring tool. With their AI Assistant, teams can build courses using natural language and existing materials like slide decks or PDFs. It drafts content and generates assessments, imagery, and narration.

Works best for: Teams already using Articulate or teams who need SCORM or xAPI-compliant output.

Falls short when: Teams have a robust SCORM content library they're looking to maintain within one platform.

4. Video platforms

L&D teams are increasingly investing in training solutions that offer personalized learning at scale, like video training. AI video platforms transform what was a time and cost-intensive production into a lightweight workflow that any instructional designer can pick up quickly.

With AI-powered platforms like Synthesia, instructional designers can direct the video creation process. They can start with a prompt, a script, or an existing document like a transcript or SOP, and produce a refined training video that can be published or distributed in the flow of work in over 160 languages.

Synthesia turns written training content into presenter-led videos, whether you're starting with a script, an SOP, or even an old slide deck. You can quickly generate a first draft of a video, which can be especially helpful for training that needs to be frequently updated and delivered at scale.

Want to see what an AI-assisted video creation workflow looks like? Watch this quick overview.

Works best for: Training that needs frequent updates or localization, process demonstrations, and content delivered as a supplement to an existing program, like onboarding, compliance, or management training.

Falls short when: You want complex scenario simulations or cinematic production quality. 

5. Coaching

When I first started my L&D career, coaching was reserved for senior or "high potential" employees. Good coaches are expensive, rightfully so, and few L&D teams have the budget to make even affordable coaching accessible to their broader employee population.

In the past decade, platforms like BetterUp have sought to change this by connecting experienced global coaches to employees through subscription models. Now, these companies are making that experience even more affordable through AI coaching.

BetterUp Grow is an AI coaching platform built on behavioral science and informed by over five million coaching sessions. It is customizable to your organization, meaning you can share guiding documents like career ladders or skills libraries and company values. That way coaching is aligned to your organization's priorities.

Works best for: Scaling practice and feedback reinforcement after live development programs and offering just-in-time support around critical moments like performance reviews.

Falls short when: You want a human coach with all the social components, advice about careers or empathy around the emotional toll of work. 

Beyond the core stack

In addition to the categories above, there's another set of tools worth considering when assessing your tech stack. AI notetakers, meeting transcription tools, knowledge management platforms, and enterprise search all fall into this bucket. These tools are often owned by IT or operations, but they're valuable data sources that can help bring learning closer to the flow of work.

Your team will likely also use a handful of lighter-weight tools, expensed on a credit card or through individual subscriptions. If your team needs infographic capabilities, for example, a tool like Venngage is worth exploring.

Regardless of which solution you're considering, I recommend thinking through the questions in the checklist below.

What does the future of AI in L&D look like?

If you've made it this far, congratulations, that was quite a trek. Before you go, here are a few nuggets from our survey worth noodling on.

While I'm not in the prediction business, I do think the data is directionally interesting. Only 47% of respondents said they think the LMS will be the backbone of their stack in three years. I understand the sentiment, but I'm skeptical of the timeline.

Too many organizations rely on their LMS for compliance tracking and content libraries that the change management required would be too great a lift for most. Sure, more agile organizations will get rid of it, or maybe never acquire one in the first place. But I don't see that changing any time soon. Remember when everyone said SCORM would die? Still waiting on that one.

That said, one insight I find genuinely valuable is that 58% of respondents believe AI gives L&D more strategic influence.

Personally, I think that's because it gives us the capability we've always needed but rarely been resourced for: to truly demonstrate the power of workplace development. To be a real partner to the business. The kind that unlocks talent and maybe even helps place the next big bet.