Action-First Learning to Drive Enterprise Performance

0:32 — Welcome and Intro

Chris: Welcome to another episode of Instructional Designers in Offices Drinking Coffee, brought to you by dominKnow — empowering L&D teams to develop, manage, and deliver impactful learning content at scale. Paul and I are both team members at dominKnow. Joining us this week is Dr. Karl Kapp, a returning guest who needs no introduction in the world of instructional design and L&D.

1:40 — Meet Karl Kapp

Karl: Great to be back. I'm a professor of instructional design and technology at Commonwealth University — formerly Bloomsburg, which consolidated with two sister schools. I primarily teach a graduate program focused on designing, developing, and delivering online instruction. I also work on games, gamification, and how to make learning impactful, meaningful, and tied to real business results.

3:05 — How AI Is Shifting L&D

Chris: AI is the backdrop of almost every conversation we have now. From your vantage point as an educator and researcher, what are you seeing change in how we approach L&D?

Karl: It's a little retro and a little forward at the same time. For years, the focus was on how much content we could create and push out. Now that AI can produce content almost instantly, the real question is: what's the differentiator? The answer is designing instruction that gets learners to practice and become proficient — not just to have knowledge. I always ask groups: how many people know they should exercise every day? Everyone raises their hand. How many actually do it every day? Nobody's hand stays up. Knowing and doing are not the same thing.

3:31 — Practice Over Content

Karl: What has been missing from training for a long time is practice. We tell learners something, assume they know it, and expect them to do it. That is not how it works. We need to create practice environments. That is where instructional designers add real value — because AI can help generate content, but it does not have the insight to set up effective practice, give situated feedback, or read human cues the way a skilled facilitator can.

Chris: We have been talking about this shift on IDIODC for years — moving from pushing information at people to creating something that results in actual behavior change, improvements tied to the organization's bottom line. If L&D can step out of the role of gathering and reformatting content, we can focus on those higher-value activities.

5:13 — Building Learning Ecosystems

Karl: Our charge now is to build learning ecosystems — what used to be called learning surrounds. One learning experience will not change behavior on its own. Research shows that even after a heart attack, people revert to old behaviors after six to eight months. That is a life-or-death situation. Think about how hard behavior change is for something like adopting a new sales model. In the age of AI, we need to create ecosystems that sustain performance improvement over time. That means moving beyond training toward genuine performance support — something researchers were advocating at the very start of my career. It has come full circle.

7:35 — AI Scenarios and Chatbots

Paul: How does AI actually remove the barriers that stopped teams from creating more practice-based and action-first learning in the past?

Karl: Two things changed. First, content creation used to take a long time — interviewing SMEs, validating accuracy, observing behavior. AI has compressed that dramatically. Second, scalability was a huge barrier, especially for branching scenarios. With traditional branching, learners would see three choices and say none of those are what I would actually do, or they would pick the longest option assuming it was correct. With AI, the learner can have a real conversation. The AI evaluates responses naturally and adapts. That is a fundamentally different kind of practice experience.

Karl: Beyond that, part of the future is not creating courses at all — it is creating chatbots and autonomous agents that coach people through processes. Think about the journeyman-apprentice model. A good journeyman could adapt, give stretch projects, and offer one-to-one feedback. When we scaled training and moved it online, we lost all of that. We took the broadcast model — an instructor in front of a thousand people — and just automated it. AI now lets us individualize at scale again. I have built a chatbot that debriefs my students after a game design exercise. With a few refinements — more specific prompts, telling it what not to say, giving it a persona — it now helps students focus more effectively than a group debrief alone could.

11:28 — Keeping Humans in the Loop

Karl: Bloom's Two Sigma study found that students tutored one-to-one scored two standard deviations above students in a traditional classroom. AI now makes that tutor available to everyone. Eventually AI will have sensors and contextual awareness — imagine it monitoring someone on a manufacturing floor and giving real-time performance cues. But no matter how sophisticated the technology gets, human judgment still matters. If we abdicate our responsibility as designers or managers in favor of AI, that is going to cause problems. The human must stay in the loop.

13:50 — Blended Learning with AI

Paul: Where does the instructional designer fit in a world of chatbots and agents? How do we make sure learners are actually developing skills rather than just looking things up?

Karl: Learning is not a monolithic thing. Learning to tie your shoe is different from learning brain surgery, which is different from learning to comfort a grieving colleague. Some learning has to be human-to-human and peer-to-peer. Some is purely informational and AI handles it well. The judgment about where each type of learning fits — and how to blend AI with human interaction — is exactly what instructional designers bring to the table. I view AI as a partner, not a replacement.

Karl: The most effective approach uses multiple specialized bots, not one bot trying to do everything. One bot assesses performance. Another looks at conversational skills. Another provides content. Another coaches. Think about when word processing first came out — everyone put 50 fonts on their newsletters. It was a disaster. The people who knew design rose to the top because they could leverage the tools better. The same thing is happening now with AI. An operations manager could theoretically build her own AI coaching team, but she is trying to run a division. She does not have time. Instructional designers can come in, apply task analysis and observational analysis, figure out what she actually needs, and orchestrate the whole system. That is real, measurable value.

18:17 — The Case for Practice

Karl: It amazes me how little practice is built into corporate training. Think about Tom Brady. By the time he got to Tampa Bay, he already knew football at an elite level. But he practiced every single day. He once said he liked practice better than the game because it made him better. We take a salesperson with 15 years of experience and say they do not need to practice — but do they have 15 years of experience, or one year repeated 15 times? That salesperson is practicing right now. They are practicing on real customers, and sometimes they get it wrong. You are losing sales you are not even counting. A life science company I worked with introduced AI practice scenarios and usage hit roughly 20 practice sessions per learner per day. Once you give people a judgment-free, personalized, targeted practice opportunity, they take advantage of it.

19:52 — Avoiding AI Slop

Paul: Every new technology creates pressure to do more with less. How do we avoid AI just giving us a faster way to produce training that nobody uses?

Karl: Who wants to learn leadership from an AI avatar that has never led anything? That is a real and growing problem. Organizations are already seeing the consequences of cutting too deep. The TSA incident where two controllers were doing the work of four — that is what happens when you cut past what the work actually requires. Some organizations replaced human roles with AI and months later were asking those people to come back. There is a continuum. On one end, worst case: AI replaces critical human judgment and things break. On the other end, best case: AI handles the routine so humans can focus on what only humans can do. We need to fight for that second scenario.

21:16 — Proving L&D Value

Karl: L&D often falls into the role of order takers — we just do what we are told. That has to change. Think about a doctor. They tell you to stop smoking even when you do not want to hear it, because their job is your health, not your comfort. L&D leaders need to operate the same way. We are in a unique position — we look at what people need to learn, what they need to do on the job, and what performance outcomes the organization needs. Nobody else is looking at all those elements simultaneously. We need to assert ourselves, push back when appropriate, and say clearly: we are not content jockeys. We are performance improvement professionals.

24:12 — Uneven AI Adoption

Karl: It is 2026 and I still encounter organizations with no online learning at all. They built something during COVID and it disappeared. AI adoption follows the same pattern — it is unevenly distributed. Some organizations will not touch AI for security reasons. Some use it only for content outlines. Some are building intelligent agents. Some are trying to automate their entire LMS with AI. William Gibson said the future is already here, it is just not evenly distributed. That is exactly where we are. The existential anxiety many L&D professionals feel is not yet fully warranted. It is foolish not to understand AI — but the urgency depends entirely on where your organization sits on that continuum. The future is always an and, not an or.

26:16 — Action-First Learning: The Book

Chris: Tell us about your new book, Action-First Learning. A lot of what we have been discussing flows directly from it.

Karl: The core idea is that we can no longer passively provide content and call it training. The book covers specific instructional approaches that get people moving — figuratively and literally. Each chapter opens with a challenge — an action I want you to do before you read. Then we look at real companies using action-first methods, explore the underlying principles, and discuss how AI can support each approach. Every chapter includes step-by-step instructions, examples, reflection questions, AI prompts, and worksheets. It is a culmination of everything I have learned from writing previous books and incorporating reader feedback over the years. Illustrator Kevin Thorn created action-figure superhero drawings for each chapter, and those illustrations are available as a free coloring book download at td.org.

27:26 — Card Games for Learning

Karl: I am seeing a pushback against purely digital learning. People are coming to me having developed card games for training and wanting feedback on them. There is a chapter in the book on using card games to facilitate thought, action, and practice. The cognitive technique transfers even when the physical tool looks different from the job. An insurance professional might never use a card game at work, but they categorize and filter claims all day long. A card game that asks them to sort cards into piles uses the exact same cognitive process. Physical manipulation of elements has a measurable impact on cognition — the research is clear on this. The tool changes; the thinking stays the same.

29:12 — Immersive Nursing Scenarios

Karl: We are developing a 360-degree video experience for nursing students. Instead of being told where to look for bacterial infection risk, students are placed inside a patient's room and asked to find it. That produces far better retention, recall, and application than a bulleted list. Bulleted lists do not exist in nature — we invented them for efficiency. Nature requires moving around, touching things, thinking actively. Putting learners into a situation and letting them discover what they do not know is far more powerful than telling them in advance.

Paul: So the principle is: lead with the situation itself rather than with explanation. Put learners into it, let the learning emerge from doing — ideally in a safe environment where failure is instructive, not harmful.

Karl: Exactly. Malcolm Knowles said adult learners engage most deeply when they know they do not know something. Passive content lets people say, yeah, I kind of know that. Dropping them into a live situation immediately surfaces the real gap. When it happens inside a game or simulation, there is psychological safety — people attribute failure to the scenario, not to themselves. That makes them more open to feedback and more willing to try again. We all remember our aha moments. If we can engineer those moments reliably, the learning sticks. The key constraint is bounding the experience appropriately — I cannot let someone spend four days failing to figure out a new ERP system. The challenge has to be scoped and achievable.

37:45 — Deliberate Practice and Feedback

Paul: Corrective feedback seems central to making action-first approaches work. How does that play out in practice?

Karl: It is not 10,000 hours of practice — it is 10,000 hours of deliberate, timely, constructive feedback. That distinction matters enormously. With AI simulations, we can provide feedback during or immediately after that says: here is what you said to the upset customer, here is why it was not effective, here is what you should have said, and here is the concept behind it. That level of specificity lets learners understand exactly what went wrong, how to correct it, and what theory supports the change. A multiple choice knowledge check will never give you that depth. Deliberately designed feedback is what drives transfer to the job.

40:34 — Personalized Learning and Flow

Karl: AI can keep each learner at the zone of proximal development — the edge of what they currently know. This creates the conditions for flow, the state Csikszentmihalyi described where challenge and skill are in balance. When training is too easy, learners disengage and say they already know this. When it is too hard, they give up. AI can dynamically scale the difficulty for each individual learner — Karl starts at basic sales scenarios, someone else starts harder. The instructional designer's job is to set the framework: the triggers, the milestones, what everyone needs to know, and how it maps to the organization. We are the architects. AI builds within the blueprint we create.

Chris: Personalized learning has been the holy grail of L&D for decades. Every new wave of technology has tried to deliver it. Maybe AI is finally the thing that actually does.

Paul: And this creates a real opportunity for better measurement — a more concrete linkage between learning activity and actual performance improvement on the job.

42:54 — Measuring Performance and ROI

Karl: I recently wrote an article on three levels of metrics for gamified learning. Level one is engagement — learners have to be engaged. Level two is learning — we are reasonably good at measuring that. Level three is performance, and that is where we fall short. To make performance measurement work, it has to be tied to KPIs and the organization's actual performance metrics. When someone asks me for the ROI of action-first learning, my first question is: what is the performance goal we are trying to meet? If meeting that goal increases sales, those increased sales are the ROI. The gap we are missing is the bridge between learning data and performance data. Estimates suggest around 80 percent of organizations do not measure the performance impact of learning — and in my experience that number may be even higher. Part of the reason is that we do not know to ask the right questions. Part of it is organizational silos — sales teams guarding their data, finance not sharing baselines. We need to push for access to that data. If you want to know whether training is effective, you need a baseline before the intervention and a measurement after it.

45:36 — ATD Conference and Where to Find Karl

Karl: I will be at the ATD conference presenting a 20-minute session on Action-First Learning — and I am running it as a choose-your-own-adventure format, where the audience votes on which direction we go. It is an experiment in live application of action-first principles. The book is available through ATD and everywhere books are sold — search Action-First Learning Kapp at td.org. The chapter illustrations are a free coloring book download. Find me on LinkedIn at linkedin.com/in/karlkapp and at karlkapp.com.

47:04 — Wrap Up

Chris: Karl, it has been a real joy having you with us. Your position at the university gives you a kind of fishbowl view of what is changing — you can see patterns that people inside organizations are too close to see. That perspective is genuinely valuable for our audience.

Karl: Thanks for having me. You guys are so much fun. Go take some action.

Chris: Thanks Karl. And thanks to everyone watching and listening. Every episode of IDIODC is brought to you by dominKnow — helping teams develop, manage, and distribute L&D content efficiently at scale. Visit dominknow.com. Join us again in two Wednesdays.