Episode Transcript
[00:00:00] Good morning, everyone. Today I want to give you some perspectives on where AI is going as an architecture. Because so many of our business issues and investment strategies are revolving around AI, we need to kind of have a sense of where we're trying to get to here. So I'm not going to talk in too much technology, but a little bit. So if you go back in time and sort of think about computing for the last couple of decades, the big transformation that took place during most of our careers was the cloud. And the cloud was a way of taking computers and data systems, putting them into big data centers and building tools and interfaces so that vendors or corporate people or individuals could build applications and deploy them very quickly without having to license software on CD ROMs, which by the way, is the way it used to work and install them in your data centers. And so it happened starting in the early 90s, late 90s, early 2000s was all of the corporate applications and tools started going to the cloud and the big ERPs, as SAP, Oracle, Workday, all the others went to the cloud. And we bought applications on the cloud and then we as companies stitched those applications together with tools like Microsoft Teams or other front end tools that made it easier for customer, for employees rather or customers to get access to our systems. So we built all sorts of websites and interfaces on top of this, including mobile apps. And the mobile generation that came through the cloud was in some sense a layer on top of the cloud where we could build an app on mobile and interface with the apps on the cloud. But the cloud was kind of the fundamental big thing that happened. And the cloud vendors were growing like crazy. And all the investors invested in cloud vendors and the multiples, stock multiples of cloud vendors was high and they were the darlings of the last two decades of tech. And then AI started two and a half, three years ago and we started to get this sense that two or three things were very different. First of all, with AI, you don't have to build a user interface because you can talk to it, you can chat with it, it understands what you're trying to do and it can interpret it. And you know, that saves a massive amount of work on user experience design and user interface design. It sort of personifies the computer. So the computer is, you know, your computer now. It literally listens to you. The second is the AI is a data rich system. So it has lots and lots of data and it can answer complex questions and manipulate and integrate and interpret data and generate answers in English language or another language or data or Code or other forms, including graphic images, videos, audios and so forth, which we could never do before. And then the third thing was that it was powered by these highly specialized chips. So unlike the cloud, which had to be run in a very gigantic centralized environment and only a few companies who'd really build these big cloud environments, these chips could be put anywhere. They can be put in your phone, they can be put in your PC, they can be put in your glasses, they could be put on a piece of clothing. Actually, now we haven't seen a lot of that yet, but this is a completely different technology architecture. And the vendors of the big software systems that we buy woke up two or three years ago and said, oh my God, this is a very disruptive technology, we need to be a part of it. So if you look at SAP, Oracle, Workday, the ones that are the big ones in HR, you look at Salesforce and CRM to some degree, HubSpot, Adobe, all of the big cloud players said, okay, we're going to do AI too and we're going to create interfaces in our software that are just as intelligent using AI as you get anywhere else. But of course, architecturally they were never designed for that. So what they did was they built AI models in their cloud based systems that manipulated the data and gave you new tools. So like the LinkedIn Recruiter Assistant, which is very cool, hiring assistant, rather the tools in workday that can generate a job description based on input or you know, start to help you build a performance review, things like that. So, so the bigger incumbents who have a lot of money started to adopt AI. However, not to be undone, the entrepreneurial world said, you know, that's not such a great idea because why don't we consider AI as an agent, not a application. And an agent is something that can respond to user input, just like a regular computer, but also can manipulate other things. And so it can be sort of the master and these cloud systems can be the slaves. So this new protocol coming out that's I think going to win over, called MCC Model Context Protocol, is a way for an AI agent to send messages or transactions to another system. So, you know, if you want to book an airline flight, if you say to one of the AI tools, I need to go to New York next week, I want to leave before 7am, I want to arrive before 3.30pm, I want an aisle seat and I'm willing to spend this much money, find me the best possible airline. You can find that stuff and potentially book it and then book everything else for Your trip, which is going to be really great. And then of course a customer could do that when they're ordering products on your website. An employee could do that when they're trying to fix something or solve a problem on behalf of a customer. All our HR systems would be like that, our recruiting systems, our learning systems, all that and more or less that's getting implemented now. So you know, new recruiting tools like paradox, Maki Eightfold and others phenom can take, you know, 15 steps of recruiting and put it into one agent. And you know, you couldn't do that on the cloud because in the cloud if you wanted to implement a whole bunch of extra steps, you had to kind of redesign your app and build all the user interfaces around it. And it wasn't very extensible. So this new architecture is much more flexible to develop stuff, it's much more attractive and easy to use. The non technical consumers can approach your applications and interface with them without all sorts of fancy training. We don't have to train people how to use these systems like we did in the past. And the incumbents are sort of left stuck. And many of the AI vendors consider the big HCM types of platforms or CRM platforms is just dumb databases. And if you look at the architectural pictures that I've shown at some of the tech conferences and also coming from a lot of the AI vendors, they look at an Oracle, an SAP, a workday, a salesforce as just a dumb database. Now those systems have a lot of application logic in them and you've embedded your company's business practices into those things, including LMSs by the way, and ATS too. So they're not really dumb, but they're kind of legacy. So looking at the stock market now, those companies valuations are relatively flattening. They're not growing at near the rate of the magnificent seven, the big seven AI companies. And we don't know who the next generation of AI companies are going to be. There's going to be a new workday, there's going to be a new SAP, there's going to be a new cornerstone, there's going to be a new salesforce. Now they don't admit that and they're not going to give up easy, but they will. And as with all the other technology disruptions that I've been involved in, we don't know who the winners are going to be because these markets are very complex. And it isn't just tech defines your success. You need to know the market, you need to know how to sell and market your product and have the right investors and have a lot of luck. So what does this mean for us as HR professionals, tech professionals? Well, it means a couple things. First of all, don't assume that your incumbent vendors are going to be the solutions to the next generation of tech. They're going to try for sure. And of course, by the way, the new entrant into all this is Microsoft, because Microsoft, because of their massive footprint, is unleashing AI tools that can be connected to everything. And Microsoft is a tools vendor, they sell tools. So you know you've got a bunch of options now for building solutions independent of the old incumbent vendors you have. So, so that's number one. Number two, and this is the most significant of all, and I talked a lot about this in the Middle east last week, you are now a builder of applications. You can build a prompt in English that does incredibly complex things that would have taken you weeks of coding, if you even could code it. And you can take that prompt and you can put it into your LLM like you can put it into Galileo and you can give it to everybody. And each individual can use the prompt and it'll perform differently for them based on the data set that they have in their environment. So you are now an application developer. When you build a course, for example, or a learning experience in Galileo, learn, you upload a bunch of content around the topic. You can build a template and a style guide, an instructional template, and that is essentially a complex prompt. And it will build a course and you can literally say to the system, I want it to be a one hour course, a 30 minute course. I want it to be a podcast. And so you're building a repeatable element of code that other people can use. So now, instead of waiting for the vendor to do what you want, which never quite happens, or waiting for the IT department to fix something, you do it yourself. And this is going to unleash massive usage and application value and problem solving in your hands. So don't just think of AI as a really cool way to ask questions and get answers. It's a way for you to take the things that you know how to do and build them into repeatable prompts. And there'll be all sorts of tools to do this. By the way. They're just now coming out to share solutions, innovations, or repeatable applications, whatever you want to call them, with others. Number three, I am pretty sure that within a year or two, every employee in every company is going to have their own LLM. It's going to be on their computer. If they're a white collar worker, it's going to be in the store. If they're a retail worker, it's going to be in the cash register. If they're working in a grocery store or someplace like that, it's going to be embedded into everything and eventually our phones. So the manipulation and the interpretation of this data and the actual application code is going to run locally as well as remotely. So that implies that the stuff you build in the corporate world can be mixed and intermatched with the local information in that particular domain. So if I'm a manufacturing plant and I get a safety and compliance rule from corporate, or we have a fire and we change our policies, or we have some changes to the work schedule or whatever that comes down from corporate, we all hear about it, we all implement it, but then the local manager says, but here, because we have a water shortage, we're going to stay closed a couple of extra days and we're going to do this, that and the other thing that's a little bit different from corporate because of our situation locally, the individuals in that plant will see the local version of the corporate policy. And if the local plant people are given the ability to sort of tweak the interfaces, the HR systems that they use, they can build little code of their own that localizes the content that came from corporate. So this is a massive change in training, knowledge management, leadership development and so forth. So we end up with going back to, in a sense, a new version of the client server architecture, which was the architecture I originally was working in in sybase in the 1990s, early 90s, which was the pioneering architecture brought about by the PC.
[00:11:30] So very, very different competing systems. Now we're in the middle of a lot of architectural research, by the way, so stay tuned for a big survey that's going to come out on this. But the implications of this are really tremendous. A lot of vendor disruption, for sure, a lot of vendor reinvention, a lot of innovation. But even more so for you as an HR person or a tech person, you are a builder now. You're a creator. You can build stuff, you can develop things, you can solve problems and share your solutions. You have to learn how the basic AI works. But that's not that hard. I mentioned last week that the IATA team, who does airline industry analysis for the entire airline industry, is using Galileo and they took a bunch of data from Boeing and Airbus and other places and they built a whole new skills model for the airline industry. Every single role the skills today, the skills in the future, the growth rate of that role matched to the geography, because the geography distribution of skills and roles is different, different depending on different parts of the world. They built that in Galileo, they put it into a report, but presumably they could just send you the prompt that they used and you, as an airline executive or an airline HR person, could take that and manipulate it locally relative to your company. In other words, the IATA model could be the core corpus, and then your company's org structure could be laid on top of it and you could map it against the IATA model and see how much change you're going to see over the next five to 10 years based on the Eotta model. If you tried to code that by hand, it's going to take you two years just to code it, and you never really quite get it right. And nobody would know how to use it. And there will be many, many, many applications like this. The biggest ones will be in learning and development, which I think is going to be called enablement. I think learning and development is going to be a part of enablement. And we are now working on this with clients, on a whole bunch of applications. The big one that I want to highlight that we're launching in a couple weeks is for leadership, leadership development, leaders, supervisory support. So most of you know we have this tool called Galileo, which is a hugely successful HR tool that does a whole variety of things, hundreds of use cases, lots and lots of core data inside of it. And what we're doing is we've built a version of Galileo coming out. The next release is called Galileo for Managers, and it's a version of Galileo that takes all the data that we've amassed over the years about HR practices and embeds it in a experience that supports a manager. So if I'm having a difficult conversation, if I have an underperforming employee, if I'm doing hiring, if I'm trying to figure out pay, if I'm trying to figure out how to run a team meeting, if I'm trying to build a development plan, if I'm just struggling with my own role and what skills I need and what development I need, it knows how to do all of that stuff because we've been teaching HR people how to do all this stuff for a long time. And we have probably 500 case studies in there of examples of how companies have done this. And leadership models from shl, Heydrich and Struggles and others are coming. So this manager tool in Galileo, it's going to be available in A couple weeks. And in the sort of argument about the two tier architecture, the learning platform, Galileo Learn, which is an AI powered native content platform, can be coupled to this. So you could deliver, and we hope you will, Galileo as a management tool, a management coach, a management training system to your employees. Build corporate content in Galileo Learn and the employees and supervisors could use this dashboard, in a sense, intelligent dashboard for all of their management inquiries and needs. And then of course, you load your own policies, your own hiring practices, integrations to your ATS for hiring and other things. And you've got a high powered management support enabler development system for every manager in your company that brings the LLM technology right to their desktop. Now, we haven't decided how to brand that whole experience, but Galileo for managers is available in a couple weeks. All of those of you who have Galileo are going to have it and you can click on the button and use it and you'll see that it's all a power of Galileo with the manager layer on top of it. And this two tier architecture is very powerful because if you deploy a system like this, you can localize learning or management tools to each department, store, region, geography function very, very easily. You could never do that in leadership development before. Leadership development or leadership enablement was always kind of a corporate training function and people had to take courses or browse videos. We can now give people personalized support training, education tools for management and supervision right at their desks, right at their kiosks, right at their nursing station with the LLM in an integrated two tier architecture. Now there's a hundred other examples. Think about a salesperson, think about a manufacturing person, think about a truck driver, think about a retail worker. I mean, this architecture is really useful in companies because every single company has this problem of central versus local. The central corporate folks want to build solutions and support systems that address the whole company for scale. But one region, one division, one geography needs that translated to a different language or needs to apply local conditions or legal regulations or other, maybe even product changes around it. So we've got in the AI world a systems architecture that is extensively different and very programmable and scalable in a way we've never seen before. So anyway, that's a little bit of thought on that topic. We'll be launching the new version of Galileo, codenamed Venus, in about two or three weeks. We'll give you a lot of information about that when it comes out. Those of you that want to get to know our stuff, you can get Galileo Learn and Galileo together. Now As a bundle. Everybody who gets Galileo Learn gets the creator part, not only the consumption part. There are 700 courses in Galileo Learn, but there's also an entire creator environment. So you're getting essentially an entire library of courses, an entire learning management platform and a whole bunch of development tools, which would be the equivalent of Articulate plus Camtasia plus a bunch of others, all for $495 per user. It's kind of crazy this AI stuff works. And you know, because it's built on a native AI platform from Sauna, the level of functionality in Galileo and Galilearn goes up at the same rate of speed as all the other AI technologies in the market. So you'll be able to generate videos and audios and connect these two systems together extremely well. Already you can. Lots and lots of cool stuff here. I mean, I'm really excited about this as a tech person and I hope you are too. We are going to be doing more workshops in Galileo at the next couple of conferences. So if you come to Zrtech conferences, we'll do workshops for you there. If you are want to explore this, obviously contact us if you want us to do demos for you. We'll show you how this all works. And if you get your hands on Galileo on Galilearn, we have a lot of self study material in there and we're doing more webinars and workshops to teach you about this environment because I want you all to feel like you, you have power over these tools and you're not intimidated by them. Thanks everybody. That's it for now. Talk to you again soon.
