Explore the Training Technology Stack of the Future

Hello, everyone, and welcome to today's Training Industry webinar: The Tech Stack of the Future AI and Headless Infrastructure, sponsored by Administrate. I am Avery Vogt, Events Coordinator at Training Industry, and I'm so happy you could join us before we get started, I would like to quickly go over a few housekeeping items to help you interact with our speaker and get the most out of today's program. Throughout today's event, please feel free to chat comments into the chat window and submit questions in the Q&A window. We will address all comments and questions throughout the event or at the end of the program during our Q&A section. We also encourage you to share the information you received today with your colleagues and network via social media. Please follow @GetAdministrate and #TIWebinars so we're able to to track your contribution to the conversation on the program and to receive a short evaluation survey, and we would greatly welcome your feedback as always, today's webinar will be recorded and archived on trainingindustry.com and you will be receiving a follow up email from us that will include a link to the On Demand program for you to share with your team.

If this is your first webinar with us, a special welcome goes out to you. Training Industry exists to make connections among learning and development professionals. We offer tons of resources to support your role in L&D through live events like today's webinar as well as through articles on our website or magazine conferences, courses, research reports and or podcasts. Make Training Industry your go-to resource for learning solutions. Visit trainingindustry.com to learn more. Now to get us started, let me introduce you to today's speaker from Administrate: John Peebles. John is the CEO of Administrate, an innovative training management platform for enterprise learning and development operations, serving hundreds of organizations and millions of learners across the globe. Administrate enables training teams to organize plan, deliver, automate, analyze, and scale enterprise training operations all from a single system.

John is passionate about education, teamwork, technology and mental health in the workplace and often speaks on these subjects around the world. John, I'll let you take it away. All right. Thanks, Avery. Well, good afternoon to everyone. I'm really happy to be here. I am broadcasting today from Edinburgh, Scotland, so it's always fun for me to see where everybody's located in the world as we're getting going here and seeing folks dialing in from Indianapolis (go Colts — big Colts fan) and Lebanon, Beirut, one of my favorite cities in the whole world. If you want to keep the chat rolling and ask questions, I'll try to be a little bit interactive as we go through, but we'll have at least ten or 15 minutes of Q&A time at the end for you to ask questions then as well. And today's topic, I'm really excited to share with you and talk about because it kind of combines two things that I really get excited about technical stuff or nerd stuff, as my girlfriend puts it, and also things, things thinking about the future.

And I really love science fiction. I really love, you know, thinking about what's coming next in technology and so forth. And so kind of combining these together in talking about what is the text book of the future and and what we think about this in this problem and how it relates to training and learning development is something that I get really excited about. So let's dive in. When most people think about the future, they think of something like this. Right here is a flying car. I remember going down a rabbit hole about ten years ago: there's actually a guy that has spent his whole life and significant amounts of money building a flying car that looks not too dissimilar to this. And, you know, news flash, it doesn't really work very well.

And it's kind of bulky and it's hard to get in and out of the garage. And, you know, it's one of those things. Right. We were promised all of this innovation, technology, technological advancement, and instead we just got doomscrolling. Sorry to interrupt you, but you are actually not showing your slides just yet. oops. Sorry about that. We'll do that right now. There we are, the flying car. And and so this has been, you know, immortalized in Popular Mechanics, in Popular science and so forth. And we've all seen these pictures. And you know, that that joke about we were promised flying cars and all we got was doomscrolling on social media. Well, you know, that's kind of the reality. And there's a good reason for that. And Bill Gates kind of put it in a very simple way, which is flying cars are not a very efficient way to move things from one point to another.

There are many better ways to get people from places that they want to be to places they want to go and flying cars are not one of them. And so before we launch into talking about what how this applies to AI and learning and and training, I want to ask a question for us all, which is, has your organization started a conversation started a conversation about how your tech stack will support AI tools? And we've got a few choices up here for you. A) We haven't really considered what I will mean for our software. B) We started some early discussions on how to support AI tools. C) we've been seriously discussing the strategy for AI support. D) We have a comprehensive plan in place for supporting future A.I.

tools, so I feel a bit like a pub quiz master, but I'm going to read it again, which is: Has your organization started a conversation about how your tech stack will support AI tools? A) We haven't really considered. B) We've started some early discussions. C) We've been seriously discussing the strategy for AI and D) We have a comprehensive plan in place for supporting future AI tools. And Cynthia is Cynthia is absolutely right. I was just testing everyone to see if people were paying attention on on the screen share. So take off take a minute to think about that. Answer the question when you get get a chance. But I want to move on and talk about the world's first elevator shaft.

Now a little bit of context here. This is a building that was built in 1853, and it's a beautiful old building. And it was actually built by a guy named Peter Cooper, who I heard about before we started preparing for this, the seminar. But he's an absolutely incredible guy. You should totally check him out. We don't have time to go through all of his various achievements and things, but there's two things that really stuck out to me. One, he's got one of the world's one of the most epic beards of all time. So if you see a picture of him, it is just absolutely tremendous, this beard of his. And two, he was a prolific inventor and also somebody who's really dedicated to the cause of equality and nondiscrimination and so forth. And as as he was famous for inventing and building the first American steam engine known as Tom Thumb, he helped build the first transatlantic telegraph, cable and a whole bunch of other things, including Jell-O.

So, I mean, this guy is is very important in our lives. But he also did this where he built out an adult education focused university that needed a few buildings from which to operate. And this is one of them. And it's called the Foundation Building. And actually, this building has become very important in American history. It's something like ten or 12 presidents have given really, really important addresses during whether the kicking off their campaign or active presidents, including Abraham Lincoln and so forth, all the way up to Barack Obama and Bill Clinton. And this building has seen a lot of history and it's been involved with a lot of things. But it was also the world's first elevator shaft. And the building was built in 1853. And the elevator shaft was part of that construction. There is only one weird, strange thing about this, and that is the first passenger elevator did not get installed anywhere in the world until 1857, four years later.

Now, what does that mean? What what was very interesting about this is Peter Cooper, who's a prolific inventor and a guy who really understood science and so forth. He was convinced that as buildings were getting taller, that elevators would become required and somebody would have to invent them. And so he didn't have time to invent the elevator himself, although he probably could have. And instead he just directed the architect to put an elevator shaft in. And when elevators were finally invented, they would install one. And that's exactly what they did. Somebody was going to invent a safe elevator and Peter Cooper was going to use it in his building when it arrived. And this is exactly kind of where we're at in this revolution in that most of us on this on this webinar, we will not necessarily have fully formed thoughts around what the strategy for a company is going to be, or if we do, they may change over the next three to 5 to 10 years.

But it's a big deal. There's a lot of AI coming and we need to get ready for it. It is a certainty that I will impact businesses, but we don't know exactly how. So we need to get prepared. Right. And so what does this mean for predicting the future of AI training tools? Well, you get out what you put in right? AI tools can help us process a lot of data way more efficiently than people, but at the end of the day, they're still just operating on the data they have access to, right? So, you know, you could make my brain a million times smarter, but if I haven't seen the data or haven't consumed the data, I'm not going to get to answer your question correctly. And the other point I think that's really important is back to that flying car. We can dream up flying car use cases for AI. Like all of a sudden we won't need to hire anybody ever again and it will reduce workforces by 99%, all this stuff.

But that's kind of a flying car use case. What happens when we actually focus a bit more on the practical expressions of AI and start incorporating things little by little into our workflows and into our strategy and we try to plug this stuff in and power them up. And so we need to consider what will happen. And actually the sad reality for many of us is that this is kind of what will happen when we start to try to incorporate I right. What happens when we try to plug in complex, data driven AI into the current typical learning stack? Well, we've got some gears being sheared off here. And the reality is most of the time this application is not going to work with our current data layout and our current tech stacks because things are disconnected.

We have maybe custom systems in play and it's difficult to get all the data into one spot. And even if we have it all in one spot, making sure that it's well formatted, structured, sanitized, etc., that's really important. And that will be where the the struggle is and really lack of integration, support is going to be one of the key barriers that will prevent organizations from incorporating this game-changing and really, really meaningful AI. That's going to be going to be useful for us on into the future because we'll have we have poorly integrated platforms. The data is spread out, it's siloed. You can't get in one place. It's really, really impossible to support the type of complex analysis that will need to to feed these applications.

And data access becomes a challenge, not a force multiplier. All right. And you know, let's talk about an example of this. Building, an LLM chat bot that can answer questions about my data. Now, LLM, of course, stands for Large Language Model, and many of you have heard of Chat GPT and many of you have played with it, and maybe it's on your phone right now and maybe it's, you know, writing your homework assignment or whatever as we speak. That's all fine. But many of us have this have this idea of, look, we've got a lot of data out there and we want to be able to interrogate it. We want to ask questions of it and get answers back in. If we were if we were doing this talk live, I would ask for a show of hands in the audience.

I'd say, How many people here have ever played the game Dungeons and Dragons, Right? And what normally happens when I ask this question is a single person and it's usually a guy will go, Woo hoo, right? And make a big noise about that. And it'll be dead silence around the audience for the rest of the room. But then what happens is at the end of the talk, if we're milling about in this, we're going to a real life event. Many of you would come up to me and kind of say, I've got a confession: I play Dungeons Dragons as well. Right? Or, you know, but it's with my son or some sort of excuse like that. And it's okay, everybody, you can play. Dungeons Dragons is a great game. It's got this huge resurgence and so forth. And one of the favorite things about my job here in Administrate is I get the chance to fly all over the world and talk to our many customers and visit them and see how our software is helping them and how we can improve our software and so forth.

And it's a huge part of my job visiting customers and visiting with prospects. But what that means is I don't have as much time as I would like to be able to devote to Dungeons and Dragons. Right? And if you've ever played Dungeons and Dragons before, you'll know that there's a lot of campaign data that can be generated as you're going through your adventure and your worldbuilding and so forth. And I was thinking to myself, Wouldn't it be great if I could get a chat bot to actually ingest all the campaign information from this campaign that I'm running and then I could just answer it, ask it questions sometimes, maybe even sneakily during the game, like, what did we do, you know, six months ago in the campaign? And it could answer, right? And actually I went to our product and engineering folks and I said, Hey, I've got this great idea. I want us to invest. I want us to invest R&D resources to build a chat bot to help my Dungeon Dragons game.

And they said, no, absolutely not. And so I said, well, this could be really practical because it's the exact same thing that many customers want to know, where they want to feed all their training information and all their training data and in to get answers back and maybe have a create content for them that's relevant and so forth. And the R&D team told me, no, again. So I said, you know what, R&D team? I've got my own computer science degree. I had a specialism, as we would say, in the UK in AI that I graduated with back in 2003. I'm going to do this myself. Okay. So I did right. And I want to explain to you how I built an LLM chat bot that can answer questions about my data because I think it's very relevant about not only to the things that we want to be doing as learning and training professionals, but also it gets across an important point that I learned, okay, but let's let's talk about how you do this.

Right. It's actually very straightforward, okay? And that is you've got a bunch of different types of content, right? You have SCORM content. We have training content in the form of PDFs or websites or whatnot. With video, we have other materials like Word documents and so forth. And basically we try to get transcripts of all this stuff and get it into a form that we can pass and we run it through a parser. And what we do is we split up the data into chunks. Okay, Now this this is pretty straightforward, right? You want to keep your chunks about 4000 to 8000 characters in length, and you want to make sure that your chunks overlap, right? So you actually want about a third of chunk one to overlap with about a third of chunk two and the same thing for chunk to overlap in chunk three, right?

And the reason that you want that is because you want to make sure that you can get at the essence of the answer when you are asking a question. Right. And what you do is you take these chunks and you embed them and you generate embeddings using a vector database. Now, we don't have to get into this. You don't really need to understand this, but it's kind of cool. And what a vector databases is basically it is a vector or a direction, right? A line, in space, in math and basically the line for, you know, a phrase that answers a question or a piece of data might go up in, you know, the slope is this and that. And then a similar question and phrase will have a similar vector, right? And so we we put this into a database and there's open source stuff to do this.

And it's not as complicated as it sounds. And then we're kind of ready to go, right? And this is how it works. So what happens is we get a query and we say, you know, at what point did the goblin attack? You know, the the Paladin in the campaign? Or we have a query that says, you know, what is the best way to fiberglass the whole of this model of both that we produce and our factory. Right. And what happens is we first search our vector database and we find an embedding that matches, right? And we say, okay, here's a little chunk out of the fiberglass manual or out of the Dungeons Dragons journal that talked about when we killed the Goblin and it feeds that. It takes that embedding and it feeds it to the LLM.

It could be ChatGPT, It could be any number of these models. And it says basically, Hi, you're a bot. And we want you to answer the question, which is this query using this data, right, that we are feeding to you as part of the prompt. And what happens is the LLM is trained on responses and it takes the embedding that you give it and it it passes it and it goes through it and then it basically generates the response. And this is how it works and it works pretty well, in fact. And it was really like not a not that big of a deal it was a kind of a weekend project. I got really excited about it. I sat there just knowing everybody in my life and saying, give me a query, give me a query. I'm going to feed into my bot. And everybody, you know, basically started blocking me and from texting them and so forth.

But I had a lot of fun doing this and it was a really, really fun exercise. Got to do it in Python, which is the language. I didn't spend a lot of time building in. But the the upshot was and what really, really surprised me, right, is this, which is the bottom line is your data really matters. And I don't know why this surprised me so much because if you spent any time thinking about it, this would probably be self-evident. But what I realized was as I was going through and I was kind of preparing my data and I was feeding it in, things really, really mattered about how I structured the data, about how I laid it out, how the order in which I fed it in. I had to pay attention to the chunking and making sure that the overlapping was good and what it meant was if my data was well formed and I thought about it and I built out a structure and a strategy and I fed it in that way, the answers came back were really quick.

They were really high quality. They were obviously right and in life was good. But if I just kind of took a big ball of data and didn't do anything with it and just shoved it in, which you can do, by the way, there's plenty of tools out there that will say: "Give us your data and we'll shove it into an AI thing and it'll spit out answers and so forth". What happened was the answers I would get back would be wrong, sometimes very subtly wrong, but but materially wrong. It would be hard to detect that it was wrong and and the quality would plummet. And what that meant was that trust in the whole thing would plummet. And so ultimately I found myself relearning probably the fundamental bedrock Computer Science principle, which is: garbage in, garbage out.

We may have already heard this. Many of us will be familiar with this, but it's very simple, right? If we put garbage in to the computer, it doesn't matter if it's a super sophisticated, all powerful, all singing, all dancing, AI... we are going to get garbage out. Okay. And that was one of the things that I learned. And by the way, you know, we'll send out these slides and so forth. So if you're interested, there's plenty of blog posts and things out there that talk about and this the structure and strategy of getting your own data fed into an LLM and, you know, happy to answer any questions. And like I said, we'll have Q&A time at the end. But it was a really, really interesting exercise. It's something that I will certainly use at some point and Administrate. But this idea that your data is fundamental to the quality of what is going to happen on your outputs is something that we will never escape.

It's it's just a it's a fundamental thing. Okay. And another application of AI that we've been thinking about a little bit here at Administrate. And actually I'm happy to report that our R&D team has been working on this for actually a couple of years is around scheduling. Right. And this is a problem that plagues large companies, particularly large multinational enterprises that need to deliver lots and lots of standardized training all around the world to thousands and thousands employees or partners or customers. And managing that training, which is often done in a classroom, whether it's an online classroom or a in-person class, is very, very difficult, particularly at scale.

Okay? And scheduling is one of those things that teams just have to do over and over and over again. It's very, very common. It's kind of one of these immovable forces. It's always there, it's ever-present, and it has to go well because this training is expensive, these people's time are expensive and these resources in the form of instructors and classrooms and equipment and so forth are often expensive and often very scarce. Right. There's a lot of moving parts here, very, very complex task. And what that means is because there's all these moving parts and we have to do it all the time, and it's this constant, constant chore. It means that it's it's got an opportunity. There's a big opportunity there to optimize this and get massive efficiency gains. If we can only figure out a way to do that.

Right. And so one of the things that we've been talking about and thinking about and researching and we're really, really excited about is solving this problem of scheduling. Classroom training at scale is different from self-paced training because self-paced training is kind of fire and forget. Classroom-based training and blended learning require people to be at certain places at certain times all around the world and at different locations. And it's very, very difficult to do and it's very, very difficult to automate. And so what we've been working on Administrate is an application where we have what we call the Scheduler. And on the left hand side, we've got all the things that we need to feed in to, to, to our decision-making process, right where the learners are in what who are the learners that need to be trained at what dates and times of day available at what dates and times are instructors available, what locations are available, what courses do we need to run and what equipment do we need, and so forth.

And within our software we basically support the idea. Of course, templates you can define exactly who, what, when, where and why is required in order to run a class. And that's great. It saves tons of time and everybody is super, super excited when they go live with Administrate because we can solve this problem for them and automate 80 to 90% of this headache. However, the problem that we haven't been able to solve until very recently is the idea that you okay, it's great that we can schedule one course and do that very quickly and efficiently, but we actually need to build out a one year schedule or a six month schedule. And that's where we want to enlist the power and the help of AI. Right? And so what we have been doing and working on is basically taking these templates and these actors and things that we these resources that we know about and matching them up and basically having an AI-powered Scheduler lay out a one year schedule or a six-month schedule or one-quarter schedule, how you think about your schedule in terms of time and optimize it and basically then provide a canvas for you to to work from and to make adjustments as you see fit.

And it's really, really exciting. But one of the big questions here is how do we get these tools, the data that they need? Well, with Administrate and the Scheduler, the data is all there in our platform and we're ready to go. But with the idea of maybe there's other AI chat bots or there's other stuff that we want to do and solve with the AI, how are we going to get these tools that are coming very, very soon? We've got our well, how are we going to get our elevator shaft constructed in order to then put elevators that we know will be invented in the next three to 4 to 5 years? And unfortunately, it's going to be difficult. Okay. This is not an easy thing for for most of us to solve. And that's why we want to talk about how do we if data is the most important thing, how do we make sure that we're all equipped and set up to to basically feed the data that is required to these applications and to these processes that are going to consume it?

Okay: And that's why today we're going to be talking about headless architecture in addition to AI. headless architecture. Now, what is this? You know, there's all kinds of endless opportunities for puns and, you know, jokes about losing our heads and all that stuff. So I expect to see some of those in chat. But what what headless architecture is, is actually it's it's not a new thing at all. It's new in the L&D space. We haven't seen really any other vendor. There's one other vendor that is talking about this at all in the entire L&D space, which is mind-blowing when you think about it. But basically it's a design pattern that is widely used in technical and tech stacks to design and deliver really, really flexible, outstanding technical solutions and fundamentally what we mean is basically you buy a platform that is the body or the back end or your business logic, and then you can construct a consumer experience or a experienced front end experience, which is the head on your body.

Okay: And you can construct that yourself and have a lot of flexibility. And basically we're splitting the customer experience from the back end, which is done by the the headless architecture provider and the front end which can be supplied by you as as the owner of that. And Brant saying this topic is going over my head. Yes, very, very good. So what is headless architecture? Let's just restate it right? Basically you have your front end, which is the user interface. The stuff that you and I will interact with. It's your buttons and your controls and your forms and things that you fill out. And as you're navigating around the software, it's the stuff that you use and then you have your back end. And the back end is the business logic, and that is the things, the data that is stored about what we are doing and the way that you interface between the front end and the back end is the thing called an API and Application Programing Interface, right?

Okay: And it's a standardized, structured way to basically get data from a computer and then display it using the front end to a consumer. And I want to be really clear, a headless architecture is not a couple of things, right? It's not just a feature or checkbox on a piece of software that you buy, right? And it's not a plug and play process. And what I mean by that in terms of the feature set is headless architecture really needs to be built or architected from the ground up. Okay? You can't just go and say, Hey, are you headless or not? To a vendor, they they will have had to have thought about this for years and years and years and had to have had a very explicit technological strategy to provide a headless platform.

Okay: And, you know, this this this architecture, by the way, it's used in a lot of different industries of CRM. Headless CRM has become big or last 5 to 6 years. Headless eCommerce has become very big over the last few years. And we believe that Administrate that headless architecture for learning and development and training industry is going to become huge over the next few years, driven in part by the AI revolution, but also because there are numerous other benefits. But it's not a tool or feature that you can just bolt on to an existing platform and it is not a plug and play process, right? So this is not you need engineers to to embrace this strategy, you know, And so that's just that's just a key thing. It's not you're not going to go out there and drag and drop and build out of gooey and all this stuff for your users.

Okay: It's going to require engineering investment. But the payoff can be huge, right? And so it's like great headless architecture. Okay, John, that's great. But. But why bother? Okay, Now we we believe here to administrate that decision, support tools are the future of business operations, right? So much so that we have this kind of vision document that's kicking around within our company. And you know, it's is written about five years ago, six years ago, when we embarked on our headless architecture strategy with our platform. And there's this cheesy anecdote that I wrote, and it's basically, you know, an L&D leader getting up and eating breakfast and driving in the office, and she's conversing with this chat bot with this bot, right? Kind of like Jarvis and Iron Man and she's asking questions and it's telling her, "Hey, you know, we're running out of classrooms over over in Egypt and we need to make sure that we get some more" and things like that. Right.

Okay: And the idea is we strongly believe that AI is never going to be the end all, be all solution that is going to completely run our businesses or our households or whatnot. But we do believe that decision support, i.e. the computer or the AI helping you make better decisions is going to be fundamental. And the future of business operations. So just like we talked about with that scheduling operation that plagues a lot of training and teams, we believe that the computer can help with that. It won't get you all the way, but it'll get you 90% to a solution that then can be worked through to to completion by humans. All right. And really, if data is so important, we need to make sure that we are really investing in an architecture that supports this need for data and this consumption of data.

Okay: And we also need to make sure that we support all the other tools that are generating data within our ecosystem that might be disconnected or siloed today. Okay. And so we need a spot where you can put all this stuff so we can then feed it in to these apps that are coming down the line in the future. And really when you think about decision support based on data that that you already have, we can really optimize some major decisions and really help with day to day management. And that's the type of thing that we think a lot about here in Administrate. And I realize this isn't as exciting as a flying car or some AI VR headset, but we think that this can be much more impactful and and much more important to your operation. And wouldn't you just love to have some super intelligent MBA, you know, running around mining your data all the time, thinking up ways of how you could optimize your business in your organization?

Okay: That's what we kind of view as the future of AI. Okay. Another way to think about this with headless architecture is there's an opportunity now to increase the learning surface area. And what I mean by this is, okay, we've all logged into an LMS, we're now XP and we've sat at our desk and we've taken a course and we've all maybe done that on our phones or tablets and that's been really great. And that's a that's a big change. We've also gone to classrooms and so forth. Maybe some of us have used VR and AR headsets, maybe some of us have gone to special simulators and things. One of my favorites is our customer, Maersk out in Denmark. They have these huge simulators that are the bridge of an entire ship, right? You can walk in and the all the TVs, all the windows are TV screens and they can simulate, you know, driving this ship and or maybe sailing the ship, who knows?

Okay: You ship around. Right. And it's very, very cool. Right. And these are all examples of learning surface area. And with our piloting. Thanks, Scott. Bailing me out here, piloting the ship. And so these are all examples, though, of learning surface area and with headless architecture, you've now got the opportunity to increase this and you might increase it in different ways. And we'll talk about a couple of examples of our customers having done this. But, you know, if you if your organization today sells a product, maybe it's a software product or maybe it's a hardware product, wouldn't it be awesome if you could surface learning within the product, whatever that would be? Maybe you've got, you know, a piece of software and it can pop up on the side and explain, "Hey, maybe you might be interested in this class because you're in this area of the product".

Okay: Maybe it's a piece of hardware that you want to make really sure that your customer has been trained before they can access the features of that piece of hardware. You can do that with the headless architecture because you can call in to the APIs that you have at your disposal and you can incorporate that and build out a learning experience that makes sense for you and your organization. And so really in terms of data access, right, one of the primary issues that are preventing great implementations is bad data and poor data access. And, you know, headless architecture can provide a unified back end. And so once you get all of the data into a headless architecture, you can access it in one way the same way every time and feed it out to as many different places that you want in the centralized aspect really means that you're reducing effort.

Okay: And yeah, Leah, great question. Isn't this WalkMe and what Whatfix and all these companies do? Yes that's true but in order to use those right you have to use WalkMe or Whatfix and if they don't work within whatever application you have or within whatever product that you that you have, maybe it won't work on a phone or it won't work on the device that you're manufacturing or you're building, then you're kind of out of luck, right? You're at the you're at the mercy of the vendor and their decisions about where they want to support that learning surface area. With the headless architecture, you have the control of where that learning surface area appears to your learners, if that makes sense. You know, decision support. So beyond generative API, decision making is going to be really key.

Okay: Driving these high quality decisions is going to only be possible if our tools are the highest quality and the results are going to be staggering. And we're really, really passionate about this topic and we think it's going to be it's going to be a game changer. You also get flexibility with headless architecture, so headless ness and then maybe a new term here, headless ness minimizes the repeat work of integrating new tools. Shared data structure is really great because everything can communicate in the same standardized way and you know, you can do things like you can say, today we're using this e-commerce thing to sell our training, and today we're using this assessment tool to send out surveys to our learners, but we don't have to be locked in to that software.

Okay: And maybe we decide in three or four years we want to swap one of those out where you can do that very easily, easily. With a headless architecture you don't have to worry about lock in and maybe the vendor goes bankrupt or they say no to a feature that you really, really need. You can solve that problem. You're not at the mercy of vendors basically telling you that this is going to be how your learner experiences and you can't change it. Okay. And Composability is probably one of the most interesting pieces of this idea, and that is with headlessness, you can construct basically the UX out of whatever parts you want or need, right? Because you have your data and your back end and your processes and all that disconnect ID you can mix and match tools so maybe you want to use WalkMe and you want to use some other e-commerce platform that you really like to check our process with.

Okay: And maybe you've got an LMS that you really love and some of the features that it can bring to the table. You can use all of those together and you can mix and match and you can build out something that is truly unique and truly fits the needs of your organization. And, and it's really, really powerful and, and and it gets very exciting. And I want to give you an example of this with our customer ForgeRock. So if you've never heard of ForgeRock, they are a really interesting company. They started, of all places in Norway and they had these co-founders that said, you know what? There's this open source identity management software that's going to become huge and we're going to use we're going to build a company around it and and go for it. And this basically the the tech that they were talking about was a single sign on piece of tech right?

Okay: So we've all use single sign on. Maybe we were signing with Google, we were signing in with something that work for draw. It provides large businesses a single sign on infrastructure to to basically keep their identity management and security together. And so this company started and they came to administrate as a customer about seven or eight years ago, right. When they had just raised their series a piece of funding. And we've been with them on their journey, their growth journey every step of the way. And they went public about two years ago and have since been taken private again. And they actually provide all the back end security and identity management for the Open Banking initiative here in the UK. So if you've ever performed a banking transaction in the UK, you've somehow touched a ForgeRock piece of software and device.

Okay: All right. So it's a really incredible company. They've got a great team and they're very aggressive about innovation, particularly when it comes to learning, because what they realized was as sold this complex software to their customers, the more training that they could deliver and the better they could train their users. Not only would they buy more software, but they would embed it more and more into their day to day operations and become stickier. And so they had a problem. Right. And actually, we've got about ten case studies of ForgeRock, and we've won a few awards together with them. And basically they're there. Kevin over there, who's the VP over there, he's famous for kind of calling us up and saying, I've got an idea.

Okay: Right? And I always get very excited about this because it usually means something cool is going to happen. And sometimes our product engineering teams are like, no, there's going to be a lot of work behind this, isn't there? But we love it because ForgeRock has been behind some of the some of the more interesting features we've developed over the years. But they came to us about a year ago here and a bit ago, and they said, We've got an idea. We've got this idea we want to take all of our learning content and we want to chop it up into bite sized pieces. We've all heard this before, haven't we? And then we want to put all of that out in the Internet for free, or at least most of it out on the Internet for free. And we want it to be indexed all by Google. We want it to be tagged. We want all this complex taxonomy and we want it so that with if somebody starts to play a video or take a course and learn that they don't even have to be logged in yet.

Okay: Right. And we want them to be to learn maybe for a day, maybe for a few hours, maybe for a few weeks, and they can come back to the site and keep learning. Keep learning. But one day they'll decide to create an account and log in. And at that point, we want all the history of all of their learning that they've been doing anonymously. Right. Or at least not logged in to be married up. And they're when they log in. And that's that's our idea. That's what we want. And we are like, wow, So let us get this straight. You want to put all your content out there. You want to completely change how our user portal and the learning journey works. And you want this when like in five years. And he's like, No, I need it for like, you know, 3 to 4 months, right? And so we said, Look, this is not going to be on our roadmap, right?

Okay: Not not short term, probably not long term. Right. Most of our customers want to sell their training or at the least want to know who is accessing their training. They want people to be logged in. This is an awesome idea, but it's just not something that we can do through our own application. And so Kevin said, All right, well, if it's all right, I just want to make sure that I've got a couple of engineers on my side and we'll build out the user experience ourself. And you guys don't have to do anything, just support us. If we have questions about our API. And it's like, okay, great. And we've got a really nice developer portal that engineers can log in. They only even have to log in, they can check out our API, they can run queries, they can do all this stuff. And so basically over the next two months ForgeRock's engineers and there are two or three of them built out a completely custom user experience out on their website.

Okay: If you go there today, go to ForgeRock.com go to Backstage, which is their their learning portal. You will see that you are still in ForgeRock.com you will see that it's all branded and everything looks like the rest of the site and so forth. But all the data on that screen is coming from Administrate from a headless architecture and then being presented the way ForgeRock wanted to with the learner journey that they wanted. And if you go to ForgeRock and you start learning today as an anonymous learner, you will then basically be, you know, will be tracked. Of course, you have to accept cookies and things like that. And then if you log in and create an account one day, all of your history will be will come with you. And ForgeRock won an award for this.

Okay: They increased their search ability, they increase their engagement. They like massively increased the the users that were coming through and and accessing their content. It's this huge, huge success. And I think it's one of my favorite successes of all time for them because they did all the work they designed the way they wanted to look. They built it, but they were able to do this in a very, very fast way. There's no way that most of us would be able to build out an entire public element with this complex taxonomies and all this stuff. If you didn't have the pieces, the API, the platform to back that. So they built the very thin veneer of what you see and what you interact with. And it's the power of Administrate under the hood providing all the functionality. And it was a really big success and that team deserved all the accolades they got and they won an award for that, that innovation.

Okay: And it's one of those things where, you know, now I'm kind of like, why didn't I think of that? Why don't we think of that? That's an incredibly powerful idea, but it's never seen it done before. And it wasn't something that we could we could facilitate short term in terms of building it into our application. But a customer went out there and did that and we actually have a number of different stories of customers doing this. They want to launch an app on the iPhone or Android in the App Store. Great. They will be able to do that very, very quickly using our infrastructure, using our headless architecture. They can focus on the things that they really care about, which is the learner experience and journey, and we can focus on the things that we really care about, care about, which is that infrastructure, business, logic and data management. All right. So a few takeaways here that hopefully are coming through.

Okay: And, you know, I realize this is probably a little technical, maybe a little bit more technical than you might be used to. So happy to answer questions and so forth. At the end via email, if you can't think of it right this minute. But you know, there's already this explosion of tools powered by AI in the L&D space. It's not it's not that we're pretty sure elevators are coming in a few years. It's they're here, right. We've already seen this over the last 6 to 9 months and it's just going to continue. And we can all dream about these flying car examples of AI. And it's super fun, but we really need to actually plan for how we're going to gather and process the data that these tools require. Let's not wait around to make sure that we've got our foundations solid as these tools pick up steam and so forth. Let's let's invest now and make sure that we can solve this problem.

Okay: And then headless architecture and flexible API-based integrations are the solution to this problem of feeding these very hungry, all-powerful AI tools, the data that they need. All right. So three takeaways, hopefully that you managed to get from today's session and, you know, think back to Peter Cooper and this incredible building and his incredible legacy. This guy really understood technology. He was an innovator. He really understood trends and he spent the time to build out basically one of the first universities in the world, or at least in the United States, that was tuition-free. And it was tuition-free until just a few years ago. And it was they were specifically prohibited from discriminating based on race or any other mechanism, which was very, very unusual at that time.

Okay: And so if you were a poor kid that didn't have money in, you could you could get accepted to this university and get an education. And the foundation that he put in extended not only from the opportunities that he was giving these people, but also to the way that he built these buildings. And this building was the world's first with an elevator shaft, but did not have an elevator for at least 4 to 5 more years. And I think it's an amazing story. The only thing he got wrong, by the way, is he was pretty convinced that elevators would be round as in circular. And so the tube, the elevator shaft in this building is circular. It's not a square like we normally see today. But having said that, I was just in London the other day and there was some oddly shaped elevators in that city.

Okay: Let me tell you, particularly to American eyes. So anyway, Peter Cooper, check them out. Check out his beard. He's got incredible legacy. And this building is someplace that I definitely want to visit next time I'm in New York City. Okay. So before we hit the Q&A, I'm sure that will be interesting. Here are my details. Right? This is my email address. This is really my email address. I give this out and every time I speak and always, always excited to get emails and questions from folks. So feel free to connect with me on LinkedIn as well. And then we've got a couple of resources here. There's QR code in a and a link that'll take you to a piece that we put together about learning analytics. So really appreciate the time today. Really excited to learn more about this alongside you all and pause here now for some questions.

Okay: All right, Thank you so much, John. That was awesome. Do you have some questions coming through? Our first one is from Valerie. If you have an element, how is the information in there, SCORMs, PDF, etc. extracted? Does it require an integration therefore an additional cost? Yeah, probably. Is the the answer it really depends on your LMS, it's a great question. What we have found is that a lot of LMSs today have an API. Usually it's fairly simple, right? It's designed to get courses in and out and results back and so forth. But yeah, most LMS vendors will be able to help you extract the content that you have.

Okay: The elements, particularly if it's like a PDF or whatnot. Sometimes that will be additional cost, sometimes it won't. And the key thing is for SCORM and video and stuff like that. You'll want to get a transcript basically of the the audio content that's being played through that. And you also want to make sure that you have, you know, any slide type material or any text basically that you can glean out of that content. And then that's what you start cleaning up and getting ready to feed in to to your database. Thank you. Another question. What are some signs that our team is ready to consider ahead with architecture? great question. Very good question. So what? This isn't for everybody, right?

Okay: If you if maybe you've got a smaller L&D team or smaller training team and I mean, like, you know, three, four or five people, it may not be for you. Conversely, the team of ForgeRock was a team of five people right. But what you really, really need is you need a platform that supports the headless architecture, right? So that's kind of checkbox number one. CheckBox number two is you need something that you a project that you want to want to do, right. Whether it's built out a new learner experience like ForgeRock did or launch an app or maybe you want to knit together two or three different LMSs or whatnot. We see this a lot with larger customers. They might have five or six or ten different LMSs that are in play at the same time. And you know, that's okay. Maybe maybe you want to to have something that's common across all of those.

Okay: But the last thing that you really need, so you've got you've got the architecture, the platform, you've got the project, you'll need a couple of, you know, few engineers and, you know, it doesn't have to be ongoing engineering investment. But what we have found is that larger training teams more and more are really, really benefiting from having a couple of engineers on the team. So just like we have a couple of content people right here and they're having a few engineers that the training team can access can really help in hooking up these systems and mean that that previous question of how do I extract the data from my LMS? Maybe that can be done at no cost because you've got the know how on the team to to get going. You know, the, the idea of having a few engineers that maybe they're fractional or maybe they're they're completely the training team is a trend that we're seeing accelerate.

Okay: And it's not just for building staff, it can be for hooking things up or interfacing with the business intelligence strategy within your organization. I think it's it's something that everybody should consider. Great. Thank you. Let's see, we still do a lot of classroom training. In fact, it is our most important modality. How will AI and headless platforms help outside of online self-paced learning. Yeah, a great question and by the way, I think so for all of our customers and actually this is supported by Training Industry's own survey that comes out every year, which I look forward to and should totally all check out just a plug that. But basically 80% of corporate training in the U.S.

Okay: today is still done in a classroom, whether it's a virtual classroom like this one or, you know, some some sort of physical classroom or a combination. So that modality is still the predominant way that companies are training their teams. And what we have found is it's the highest value training. So if a company really wants to make sure you know how to do your job or they really want you to learn something, they're going to they're going to shove you in a classroom and make sure that you you consume it that way. And so I think in terms of how AI is going to impact that, you know, it's kind of like this thing where it's almost like it doesn't matter how much money is spent on the classroom or the tools or the tech or anything. The thing that actually matters at the end of the day overwhelmingly is the teacher, right?

Okay: The quality of the instruction. And we all know that we know this intuitively because we all think back to high school or college or whatever, and we think that one teacher, you know, for me it was that one teacher that made me love math. And I hated math. But the guy was so good at teaching math that I just really I got it. I was really enjoying it and so forth. And so all of that is a wind up to say, I think the modality is here to stay. I think that teachers and great teaching is still the most important thing. But I think that all of the periphery around logistics and around, you know, reviewing and, you know, quizzing and all this type of stuff that is kind of not fun for teachers and, you know, takes up a lot of time and so forth. I think that stuff will be impacted quite a bit by I.

Okay: But again, it means that we have to be able to feed in the data and do so in a very high quality way because the last thing you want is some sort of review bot or a quiz bot, you know, reviewing and quizzing people on the wrong things. So that's how we that's how we think about this. This application also, we have a question from Dale. How may I affect the measurement of training effectiveness? That's a really great question. And I you know, there are plenty of people on here that could comment on this better than me. But I just think that there are so many ways that we intuitively measure engagement. And, you know, it's kind of like after the pandemic and after lockdown ended, I told the Marketing Team, I said, I'm never doing an online presentation ever again.

Okay: Right. And guess who won on that argument? As we all sit here today? But, you know, the reason is, is because, okay, you know, I can see the chat scrolling by and I can see emojis and things like that, but nothing quite beats the idea of we're in a big room. I can hear people laughing. I can tell stupid jokes and hear people groaning, right. Which is my favorite thing. And we just get all of this information that, you know, when we're on Zoom or whatever is kind of behind a curtain, if you will. And I think that that type of analysis, is something that AI will become very good at. I also think that that's a bit creepy when you think about it. Like we've all seen these proctoring things that like, you know, track you make sure your eyes are looking at the screen, all this crazy stuff.

Okay: So I think it gives me the creeps a little bit. But I think there's something to that idea that the nonverbal kind of kind of reaction that we can we can intuitively measure as humans can probably be instrumented by a machine and probably be done very well, you know, sentiment, analysis and things like that. Very, very interesting. You know, even kind of understanding stress in my voice, you know, that I think that stuff will would be really interesting. I've always kind of wondered, you know, when the when they're doing the debates and they've got like, you know, ten Republicans and ten Democrats and two independents in there and you see the little graph and, you know, somebody says like, you know, business is good. And all the Republicans, like they they go up because they like that and the Democrats go down And whatever, you kind of wonder, are we going to have that at some point for teachers and for instructors in classrooms?

Okay: Right. Or, you know, could everybody like silently hit a button that they don't understand the next, you know, then, you know, kind of like a Brazilian steakhouse where they go in the stop sign. I don't know. It's things like that, I think are what we're what we're going to be seeing a lot more of. Right. We have another question here. How long does it take? Does it usually take to set all of this up? Great, great question. So the answer is it depends rate. But I would say there's some hope here. Okay. So With Administrate, our platform, you can be up and running within 90 days if you really set your mind to it. And we've done that for a really large, complex organizations.

Okay: Some organizations want to take a little bit longer. Some organizations have done it quicker. But you know, 90 days you can get up and run on the platform. And that's kind of getting all that classroom operational automation stuff in and up and running. And then it's kind of like, all right, let's let's spend the next bit of time or block of time in to figuring out what tools we want to hook up and what the priority is. Right? And that could take another 90 days. You know, it really it really depends. But, you know, what I would say is a good rule of thumb in your planning efforts would be, you know, earmark a year for the kind of fundamental foundational phase of getting stuff into the platform, rolling out adoption, making sure your team's familiar with it and so forth.

Okay: And that will that will usually be enough. And then at that point, you've got a headless architecture that will futureproof you for the next decade, if not more. You know, for trucks. A great example of that too, bought us when the company was like 30 or 40 people and now they're thousands of people publicly traded or not publicly traded anymore, but huge company. And because they had invested in the ground level, an architecture that really worked for them and they knew they're going to grow it, they didn't have to change out as as they continued to grow. And that that's that's something we like. We like to see. Right. Well, John, I unfortunately, we are just about out of time. But do you have any last remarks you want to share with the audience before I close this out?

Okay: No, just I really enjoyed being here today. Again, email me, connect with us, talk with us. And if they're passing, understand or want clarification, let us know because we're here to provide those answers. Awesome. Well, thank you again so much for your time. I would like to invite all of you to join us at some additional Training Industry webinars this month. You can register for these programs or watch house webinars now at www.trainingindustry.com and all Training Industry webinars including this one are pre-qualified for a credit for SHRM HRC ISPI and CBT some additional resources for our learning leaders. What is CBT and it is the certified certified Professional and Training Management program that assist you in developing core competencies that will empower you to manage the future training needs of your organization.

Okay: You can participate in a virtual practicum from anywhere in the world to find out more at training industry.com/training and join 1000 of your peers online this September. For all types you can register, you will build the skills you need to drive, impact and maximize the return on investment of your own. Do your initiatives register now for just $99 to attend large sessions specifically curated to help you develop a learning strategy from the ground up and one last reminder that an evaluation survey should have popped open in another tab in your browser and we would greatly welcome your feedback. Thanks again to today's Speaker John Peebles, and our sponsor A. And thank you all for your time and attention and we hope to see you again soon.