Episode Transcript
[00:00:00] Good morning everybody. Today.
[00:00:02] I know, I feel like I'm doing this every day. I want to talk about Google's announcements this week and give you a preview of what's going to happen at our conference in a week, in a little over a week. And thank you for listening to the podcast. We did some analysis and discovered that we have more than 4 million listeners now. And this is really just astounding to me. So thank you and I obviously always want your feedback. So Google announced a whole bunch of things at their I O conference this week, one OF which is 3.5 flash, which is a new model, which is about 10 times cheaper than the current models. And the reasoning behind that, which is something I've been writing about on Substack, is the high price or cost of AI to you as individuals and us as businesses. The newest anthropic model, Opus 4.7, is two and a half times more expensive than OpenAI per token. And so everybody's burning up all their budgets on it. And of course they're now the number one revenue producing AI company.
[00:01:06] So it's working out well for them, but not for everybody else. And you know, this high cost problem is an issue for all of us because it's going to slow down innovation. Some of these AI projects don't work. I wrote an article about three or four of them that have failed. And if you're spending 50, 70, 500, $200,000 a week or a day on tokens for something that's not going to work for companies are going to stop or they're going to slow down. And that's unfortunate because this is so new, this whole space. We have to spend some time making some mistakes to get it all to work. This is the way the Internet was created. The Internet took longer to build, but a lot of the early websites had all sorts of glitches in them. So that's one of Google's things. The second thing that's huge is the AI features in search. The AI mode in Google Search has suddenly gone live. And I think if you listen to Sundar Pichai talk about it, the reason they probably did this is they feel that they're slipping behind. And so all of a sudden if you turn on the AI mode every time you do a Google search, you see an AI to talk to, which is pulling your eyeballs, of course, away from OpenAI or ChatGPT or Copilot or Anthropic. And it works pretty well because they have access to real time data in the search tree that the other guys don't have. So you can go into the Google AI and ask it a question about what's happened in the last five minutes and it'll get it, it'll find it. Any article that was published or related topic. Of course, what they're probably going to do with that is turn it into an agent. So it'll ask you, would you like us to order that for you online? Would you like us to book a flight, Would you like us to find that for you, et cetera. So they're not going away in the enterprise space. They're not really that big of a player yet and they're trying to get in there and they just rebranded all of their AI products with more enterprise names. Because the primary market for Google has been software companies, which is huge. Of course, most big software companies have massive investments, investments in Google Cloud and Google builds for developers. That's really kind of their core strength. So that's a shift to the enterprise space. And in the enterprise space where we live, of course you've got Microsoft ServiceNow, OpenAI directly, Anthropic directly. And to some degree Google and OpenAI and Anthropic have invested in big consultancies now to try to help with the forward deployed engineering. And I guess Google's going to do the same thing at some level. So anyway, it's very important we watch out for Google because Demis Hababas, who's the founder of DeepMind and really one of the most successful entrepreneurs and inventors of this technology, really does believe that we're on the frontier of the singularity. He calls it the foothills of the Singularity. Because what they've been building in DeepMind is not just a smart model, but a system that can learn anything. Because the ultimate learning machine doesn't just learn how to code, it learns anything. Anything you can throw into it a game, a video of a physical activity or whatever system you have. It can learn from that system because of the way it can test and recursively self develop itself. It's called recursive self learning. So they're very, very good at this, although they're not as competitive in the market as Anthropic at the moment. What we're going to talk about irresistible, and I'm going to spend a lot of time on this in my keynote is putting this into the context of enterprise and putting it into the context of HR. So we're going to formally launch the HR 2030 program. You'll see the blueprint. Those of you that are Clients or use Galileo will get the blueprint and I'm going to explain it in some detail and then Kathy will explain it in some detail and we can talk about it as a group. We have about 450 people coming. The conference is full. We're really excited about it. For those of you coming, you're going to have a blast. We got all sorts of good stuff going on, but let me share a few other things that have come up this week.
[00:05:05] So the main HR vendors in our market, Workday, Oracle, SAP, UKG, Dayforce, ADP, all of them are what you might call traditional SaaS companies. And if you looked at Gartner's analysis of the enterprise software industry, you'd see that there's around 4 or $500 billion of software in that space, including security software, infrastructure software, databases management software, scheduling software. I mean, there's just a ton of stuff, vertical applications, et cetera. And the average big company has, and I mean a big company with, you know, 50,000 employees or more has 140 employee facing systems. And some of them are very small. It might be a benefits application or a scheduling application or a compliance application or something, but there's a lot of them. So those of you that are in those types of companies, which is where we spend most of our time, you're looking at AI and all this beautiful news stuff and you're saying, well, wait a minute, we got a lot of stuff already running here. And you know, it's all stitched together reasonably well. Maybe not that well, but it's stitched together in some fashion and we would like to move from point A to point B. And so, you know, there's a lot of ways to think about that. One is you use AI and agents to lay on top of what you have and you slowly, incrementally chip away at the problems you have with your current environment and then little by little replace the old systems. I think that's where most companies are going and that's what most companies are doing. However, if you do that, you may end up with a new Tower of Babel or a new, what we used to call a bag of doorknobs. That's what we used to call it when I was in the software industry, where you have now a whole bunch of things that are agentic, but they don't talk to each other. So you end up with 140 old things, maybe half of them go away in 10 years and then 140 new things. But maybe it's not 140 maybe it's 500 because the agents are much more granular. So what we're trying to prevent, and I'm trying to help everybody prevent this, is, you know, going down this path again. And, you know, of course I can't stop you from buying whatever you like from a vendor. Somebody's going to come along and sell you something that's really cool and you're going to want to buy it. An interviewing tool or a coaching tool or a training tool or a pay analysis tool or whatever. But if you have an architecture and a vision of where you're trying to go, then you can sort this out. And that is what we're doing in HR 2030, is we're going to give you this vision as best we can and collect from you and with you, all of the use cases that are emerging in this space. So the future of this space is not, as Gartner says, little by little, replacing old things with new things. It's reconceiving how you want the operation to work. I don't think we're going to have any choice because forward thinking companies are going to do this in sales and operations, in marketing and manufacturing and logistics and finance. And you can't be left behind because if the things that people expect from us, like hiring the right person, quickly redeploying people, reskilling people, developing leaders, paying people fairly, improving employee experience, improving quality, improving productivity, reducing turnover, that's what we are accountable for. We can't blame the technology as an excuse and say, sorry, we'll get back to you in three weeks. When we figure out how to do that. They're not going to wait because they're using agents on their own. So the old way of doing tech transformation, which is, you know, buy a new system and then slowly, slowly, slowly replace the old ones, isn't going to be as available when everybody in the company is moving this fast. And so the perspective that I'm going to give you at our conference and I want to give you on this podcast, is this really is a once in a lifetime opportunity for you to rethink how you do things. And I don't mean just buy a new tool and turn it on and get all excited about it. That actually is not the way to do this. Companies that go out and buy all sorts of new tools because they're excited about them always come back later and say, what do we do with this mess? That that comes later. The first thing is deciding how you want the operation to work and then finding the tools and the vendors to get you there from here. And that's. And we're going to help you with that. And we can help any one of you with that. The simplest way to do this, if you're an HR person, which most of you are, some of you aren't, but, or a consultant, is to go back to the CEO or whoever the business owner is, who you're interacting with, and say, of the 50 things that's on your list for us to do for you, what's the top one or two? And it might be, if your company's growing really fast, getting the best people in the door as fast as possible and getting them up to speed and up and running. So you could take that as your mandate and build the next generation HR 2030 solution for that. And I'm telling you, the process of building an agent implementation, and you're going to see all sorts of case studies on this at the conference, is not simple. And it's not the same as buying a piece of software and training everybody how to use it. These systems are tunable and configurable and imperfect in different ways. So if you get started on one project like that, that's the right one or the most important one for your company, you'll learn a lot. And then you can decide, well, now that we get this under our belt, let's pick this other tool from this company and let's build this one ourselves and let's use this infrastructure and so forth. So that's really the big theme of what's going on here. Third thing I want to talk about is very briefly the infrastructure of AI. This is a period of time reminds me a lot of my period in the database industry where most of the big technology vendors don't want to get cut out. So you're those of you that are involved in AI stuff, you're getting presentations and demos from Microsoft, From Workday, From SAP, From Oracle, From ServiceNow, From Snowflake, From Cornerstone, from everybody. And you know, they're all great and they all look really cool because they're just so agentic and interesting. They are just, you know, different from what we've had in the past. But what we're finding out is that because it's moving so fast, most of these technology vendors are selling ahead of their skis. They're presenting demos of things that are sort of half built. They're conceived well and they're moving along, but they're not in production with hundreds of companies yet. Which means that whichever ones you pick, you're taking a bit of a risk. Now, there's a lot of ways to evaluate vendors. It may be the incumbents are the ones you're going to use because you're just already committed to them. That's fine. That's one way. It could be that you have a strategic relationship with a vendor because of a senior executive and another financial commitment that was made by them, or you might have a relationship with a consulting firm that really advises you. And this is what we do.
[00:11:57] We know some of these vendors very well and we know kind of what they're doing. And by the way, one of the reasons we know them is we're doing Galileo integration with almost all of them. And then there's other criteria. But I think we're in a period of time now where you need to vet them as best you can and ask them to show you examples of how other companies are using their platforms. And because you'll learn a lot from that, and they will learn a lot from that, and it will help you make a more informed decision. One of the companies that's coming to the conference we met with this week, we had a very kind of frank and open conversation with them this week about their relationships with different vendors. And they were kind of complaining to me and to each other that, you know, a lot of the things that were announced and demoed didn't turn into reality. The reason for that is that when you're in the software industry, which is a brutally competitive space, whatever it is, you do don't aspire to be a software company unless you're really ready to stomach it. Is if you don't have an integration or a relationship with the incumbent in a client, you. You get cut out. So a lot of these integrations and business partnerships and App Store things are sort of half built and they kind of work, but they weren't really productized because the financial incentive of the vendor that's built it is mostly just to prevent getting cut out of an account, not to make money. Because a lot of the integration work that they do is not something they sell to you, it's a feature of the product itself. So they don't have the engineering resources to build deep integrations to every one of their partners at the same time. They do them over time, and some of them get deeper based on the relationships they have. So one of the ways to test how real these integrations are is to see how many companies are using it and what is the volume of activity of people using the integration they have. So if company Says we have a connector to this, this, this and this. I mean, the next question is, okay, I'd like a list of 10 companies using those connectors, the ones that we're going to use, because I want to talk to them. And that's just because there's only so much software engineering to go around. Okay, Final thing I want to make a comment on that came out of Sundar Pichai's interview on Google. I o, you know, everybody's really, the younger, younger generation is really angry and upset about AI. And you know, I, I understand why. I don't agree that it's the right thing for them to, to do, but I understand that's the way they feel. And then there's this narrative that all the AI jobs are going to go away and there'll be more leisure time. And then I noticed that all the CEOs of the AI companies have changing their tunes because they're going to go public. So now they're going to talk about ROI instead of job loss. But anyway, that put aside this idea that AI is going to make your life easier. I don't think it's true, and I don't mean this in a negative way, but anytime we get a technology that's incredibly useful, even the spreadsheet is the best example that I could think of, or the PC itself, it doesn't make your life easier, it makes your life actually harder.
[00:14:54] Because now you have to learn how to use it, you have to learn how to manage it, you have to tweak it, you have to fix it, you have to keep up with the new version of it. Yeah, things happen faster, that's true. There's certain things that are miraculously faster, but then there's a whole bunch of overhead around it.
[00:15:10] And I think the vendors are realizing this, that more capability, more functionality, faster token generation, more code generation, better artifacts, better tools. Great, great, great, great, great. But if it isn't simple and easy to use, you hit a limit. I mean, I suppose the best example of that was the iPhone. There were a lot of mobile devices for a long time, going back to Nokia and BlackBerry and so forth, that were great. The iPhone was so easy to use that it became a trillion dollar market cap business for simplicity. So I would not listen to the claims that AI is going to make your life easier. It's not. It's going to make it more fun, it's going to make it more interesting, it's going to make it more exciting and maybe much more enriching, but it's not going to be simple because there's a lot of things to deal with here with these tools. And that's why I get back to this issue of complexity in the enterprise space. Is that as exciting as any one of these vendor offerings may be, you're going to have to live with it, you have to manage it, you're going to have to secure it, you're going to have to update it. And just like what they used to tell me at Sybase, Bob Epstein used to say this, the most expensive software you're ever going to buy is the software that's free. In other words, if the vendor or solution provider or platform that you're buying it from isn't really, really well built to be easy to use and well managed, then it's falling on your lap to take care of it. Anyway, the Google announcements are interesting. I don't think Google's going to dominate the enterprise space anytime soon, but they're getting there. And obviously Anthropic is making a huge play. Microsoft's doing a lot of cool stuff that's going to be coming out at the developer conference next week and write an article about that. They're now embedding Anthropic into the harness and I'm working on an article about all the new words in AI and all of the new language that is being foisted upon us to try to help you guys understand all this. Okay, that's it for today. I'll keep you guys up to date this week. And for those of you coming to Irresistible, you're going to have a fantastic time. We have a lot of special things planned for you. Bye for now.
