AI can now draft a module in minutes. The judgement it can't supply is deciding which on-the-job behaviour has to change, what will get in the way, and what the learner's manager should do about it. In this session we take one real course and run a human-led workflow on it, live. You decide the behaviour, the barriers and the manager's role. AI drafts the pre-work, the manager conversation guide and the follow-up messages. You review every draft against three questions before anything goes out. You leave with the prompts, the review questions and a 30-day follow-up plan you can use the same week.
Event: Human-Led AI for Instructional DesignersSpeaker: Fergal ConnollyHost: Luis Malbas
Luis Malbas:Alright. We are live. Welcome, everybody. Thank you so much for being here. Let me take a look to make sure that the room is filling up. Yes, excellent.
If everyone in the chat area can just drop in a hello and where you're logging in from... Jennifer from Illinois, thank you so much for putting that into chat. Let's see who else is going. I love doing this just to make sure that our chat room is functioning as expected. Elizabeth from Cincinnati, Ohio. Hi, Nicole. Nice to have you here in this virtual space with us.
Alright, well, I think we're just gonna get going. Before we get started with Fergal, I'm going to let you know a couple of things. With this particular platform, there is a lounge area that is open in between sessions. If you want to chat with other folks or do some networking, you can go into that lounge area. Hopefully, other folks will grab tables for you to chat with. I'm gonna do that today, so if you have any questions about TLDC or just want to have a conversation, I'll be in the lounge area in between.
I also want to mention that yesterday, we had a great start to the event. Make sure you check out the recordings here in Airmeet if you weren't here. Mike Simmons shared some frameworks that can help you strategize how to use AI in your work, which was really helpful for me. And then Hassna Ramadan submitted a pre-recorded video on using Google Flow to produce high-quality video. That was really interesting and eye-opening; I'd never actually worked with Google Flow at all, so that was very cool.
With that, I'm going to introduce you to Fergal Connolly. Fergal is an L&D veteran with over a decade of experience at tech giants like Meta, Amazon, Indeed, and Mastercard. After realizing that true behavior change happens entirely outside the classroom, he founded Multiply Transfer to help teams manage the crucial weeks before and after a program. Today he's here to argue that in an age where AI can instantly generate content, an instructional designer's true value lies in designing what happens outside the course. It's very compelling. Everyone, please welcome Fergal Connolly! With that, I'm gonna go ahead and hide myself from the screen and let you take over.
Fergal Connolly:Thank you so much, Luis. Thank you everyone for your warm welcome, and thank you for joining me. I know you're all super busy. It's absolutely fantastic to be here.
Like Luis said, my name is Fergal. Today, we're going to be talking about how to leverage AI in the role of an instructional designer. Yes, AI can create courses for us. LLMs are fantastic technology, which, when leveraged correctly, are absolutely incredible tools for the instructional designer's toolbelt. But our job hasn't really changed, no matter what's in our toolkit. Our job is to create behavior change—to help people do things differently after the courses that we design. It's always been the job, and it still is.
So we have just 50 minutes left today, and I'm very excited to get started. What I'm more excited about is what you're gonna do next. You're gonna take all this great information I share—the prompts, the playbooks, the schedules—and bring it back to your jobs to apply it immediately, right?
No, maybe not. What's actually more likely to happen is you're gonna go back to work and immediately realize that your inbox has exploded, your calendar is wedged with meetings, and someone is pestering you on Slack or Teams with "just a quick question." All of a sudden, you can't remember what your job is, never mind all the interesting things you heard today at the conference.
The interesting thing is, what's happening to us today is exactly what happens to our learners after training. We deliver the training, but very, very little happens afterwards. But that's not your fault at all. In fact, there is a whole system working against us—a system that forces us into an event-focused mindset, keeping us searching for the next tool or AI agent that's going to magically fix the problem. But the truth is, the game is rigged.
Companies spend an absolute fortune training their staff every single year, and most of it is forgotten within weeks. People join the sessions, nod along, and enjoy them, but they go back to their desks and very little changes. The money is gone, and nobody can even tell you whether it worked. This puts massive pressure on hardworking people like you to prove your worth. Because, like I said, the whole goal of L&D is to create behavior change, but unfortunately, we have very little evidence to suggest that is actually happening.
Here's what I think the causes are. When discussing effective learning design, I like to use the analogy of an iceberg. Typically, we focus on things that are easy to see and influence. We do LNA meetings, gather data, and focus on engagement. Engagement is good—it helps increase learning—but it's just easy to see. We track attendance and create lots and lots of content.
But there's so much more required to move someone to actually change their behavior. We all know that with icebergs, 10% is above the water, and 90% is below the water and much harder to see. The things below the water are what really matter when it comes to influencing behavior change. You need to dive deeper into the working environment to understand and influence these elements. I believe 80% of our training is failing because we've ignored 80% of the factors that influence behavior change. Taking this approach is doomed to failure.
The event model supports none of this. To illustrate, imagine our little learner in the work environment doing his nine-to-five job. Then he has a training event where he is completely separated from the work environment, isolated, and given a huge amount of information in a short amount of time. The model assumes the learner will transfer all of that back to the workplace. It completely ignores all the workplace systems that truly determine 70% to 80% of the behavior change outcome.
Again, I want to be clear: this is not your fault. The real villain in this story is the billion-dollar industry created to convince us that learning happens in events—workshops, webinars, Rise courses, PowerPoints, Zoom sessions. We've all been trying to create behavior change inside a system designed for information dumps. That's why our training is failing.
With AI and the rapid acceleration of technology, there's a new tool every month that can change a slide deck or a policy document into a course. AI is going to keep advancing further into course creation. The question for us as instructional designers is not "should we use AI?" We should, and most of us already are. The question is what we use it for. Are we using it to do the same thing faster, or are we orchestrating it to do the thing that actually makes a difference?
Here's the trap: If we use AI to build our courses faster, we get more courses. But if those courses don't change what people do at work, all of the course creation is wasted. It's like buying a faster car to get you to the same dead end. Play that forward a few budget cycles. If leadership keeps hearing, "Well, we made 40 courses this year instead of 20," a CEO might ask, "If AI is building these courses, then what do we need these L&D folks for?"
I don't think it will go that way. The good news is we have a huge amount of research—over a hundred years of it—telling us what actually moves the needle for behavior change. You can melt it down into about 12 key factors. Three are about the trainees, four are about the training design, and five are about organizational factors. So only four of them are about the training design itself! Most of what decides whether training works actually happens outside of the session. We are hedging our bets by only focusing on session design.
Our value as instructional designers has always been the same: the impact of good instructional design isn't what they thought of the course; it's what they did next.
Here's the deal I'm proposing: Let's let AI draft, but we are the orchestrators. We decide the behavior. We decide the barriers that are in place. We decide how the manager can support the learner. Then, we use AI to actually build the support, priming materials, and performance support around the course.
Let's look at an example: a course on "Giving Feedback that Lands." Instead of just a half-day workshop, our learner is now fully supported. They get priming communications beforehand. Their manager knows what's going on and is available to support them. We've brought in peer support. The learner has action plans, nudges, and a 30-day check-in with their manager. Everything is designed around the key behaviors we want to influence, utilizing the best practices from transfer research.
In the resources I'm sharing with you via the link in the chat, I have some basic prompts for you. First, identify the exact behavior you want to see at work. If there's no skill gap, there shouldn't be a training. Then, understand the barriers in the workplace. Different demands, competing priorities, and workloads will get in the way of learners applying what they learned. Once you identify the barriers, you can bring the people manager into the situation to help remove or reduce those barriers.
Next, have your AI draft the materials based on this brief. The AI can create a guide for the manager outlining what the team will gain and how to support the learner. It can draft a preparation prompt to create a learning action plan so learners know exactly what they'll do next. Finally, it can create a performance prompt to structure reminders and nudges when they are back at work getting ready to apply the skills.
Of course, before you send anything out, you need to use your own critical thinking. Does the output point to an actual observable behavior? Does it involve the manager? Does it arrive when a learner can actually act on it? Your AI won't have that human context; you do.
In the space of just 20 minutes with AI, we took a standard event-driven course and extended the entire process to make it much more likely the learner will actually use what they learned. We increased the scope of influence from just four training design levers to ten full support levers.
Our impact was never about how much someone liked the course. With AI, we have a fantastic opportunity to get out of the event-driven mindset and start thinking long-term about how we support people in the flow of work. I'm happy to take any questions. Luis, thanks again for having me.
Luis Malbas:Sure thing. I apologize, my internet connection dropped out for a little while there, but I do have some questions for you. Looking at the templates and examples you provided, is there a particular AI platform you feel works best with your prompts, or is it just whatever?
Fergal Connolly:I tried to keep it quite general because I don't know where everyone is in their AI journey. If you have your basic ChatGPT interface, Copilot, Gemini, or Grok, this will work in any of those provided you give it the context by defining the barriers and behaviors.
Luis Malbas:So even the free versions would be sufficient?
Fergal Connolly:Provided you have enough context limits, you should be able to get some quality outputs from them.
Luis Malbas:Great. I've got a question from Wendy Pratner: "How do you see this working for an instructional designer in a higher ed environment?"
Fergal Connolly:Great question, Wendy. The challenges and barriers in higher ed and corporate environments are very much the same. Students have a huge amount of information coming at them from all different angles. You can definitely apply these best practices by breaking things up into a more piecemeal approach—sending things at the start or end of the week, or pairing up peers within cohorts.
I think the best way it could be leveraged is if the learners themselves use it to create their own action plans. It's not just "I listen to my lecture," but "What does this mean to me? Why does it matter? What am I going to do differently knowing what I now know?" Giving the learner tools to take ownership is a massive opportunity.
Luis Malbas:I think these prompts are an opportunity to dive deeper into getting results in higher ed and might inspire people to try different things. Fergal, do you have instances where you've used these to actually create practical training production, or how did you come up with these?
Fergal Connolly:I've worked in many fast-moving tech companies in Dublin. It doesn't matter how big your L&D budget is; the challenges are the same everywhere. Generally, L&D provides training as a solution to all problems without diagnosing the workplace issues. My approach is to look at the research. I've taken pieces of that research and applied them in my workplace—trying a manager support email, seeing how people respond, tweaking it, and then adding a learner priming email.
I didn't have to ask permission to do these things—it's just email! But it requires stepping outside the training box to do something different. Who can we learn from? We can take best practices from goal setting, behavioral psychology, or marketing, and apply them to L&D.
Luis Malbas:Interesting. Tell me more about Multiply Transfer. What's the story behind your organization?
Fergal Connolly:Having been in L&D for so long, I felt the issues faced across different companies. I tried to create software that helps L&D professionals apply these research-backed best practices to every training they run, from intake to evaluation. L&D professionals are super busy, so I wanted to create something that maximizes their effectiveness without them having to go off and do all the research themselves. It's all about implementing things outside the course, influencing the workplace where the actual behavior change happens.
Luis Malbas:What do you think has been the biggest change in L&D with the advent of AI?
Fergal Connolly:Content creation, unfortunately. It's what we've become known for in L&D—we're the people who create the content. But that's a bad road to go down. We should be known as the people who help employees perform more effectively and change their behaviors. Content alone will never get people towards behavior change; if it was enough, we already have Google for that. Content is just a spark.
With AI coming in, it has made it easy to do the easy thing: creating content. That's a cycle we want to get out of. We need to leverage AI to move away from pure content creation and get into the behavior change space as soon as possible.
Luis Malbas:I am going to clip that and post it because that is good stuff, Fergal. Thank you so much for participating and sharing with us.
Fergal Connolly:My pleasure. Thank you all for joining. Thanks so much, Luis, and to the community for setting this up.
Luis Malbas:Alright, everyone. It's ten to nine here Pacific time, so we can wrap things up. Our next session starts at 10:00 AM Pacific time (1:00 PM Eastern) with Theodora Amoa, who is going to be talking about using AI for media asset creation. I'm looking forward to it. Everyone, bookmark Fergal's page; it's really interesting. I'm gonna end the session and I'll pop into the lounge in a bit after I grab another cup of coffee. Bye, everyone!
