Episode Transcript
[00:00:00] Speaker A: Good morning, everyone.
[00:00:01] Speaker B: Today we are publishing a detailed article on the reinvention of Workday. And it's a very, very big, long, interesting, important story, but let me try to highlight it in 15 or 20 minutes for those of you that would like to listen. So Workday is a pioneering company, or it was, and really invented the first cloud native architecture for corporate applications. In the early days of Workday, everybody
[00:00:28] Speaker A: was behind except them.
[00:00:30] Speaker B: SAP, Oracle, Salesforce, all of the application vendors that had come from client server architectures had been selling SaaS, which is basically hosted versions of their software. And Workday started from scratch. And the object oriented database and the rules layer and the security framework and all the things they built were designed for scale and security and cloud architectures, which were barely being formed at the time. And so.
And they built an architecture where for at least many years they didn't acquire anybody. They built it all themselves. So all of the functional components of the HCM system and then later the financial system were all built by Workday.
[00:01:13] Speaker A: So it was as if I used
[00:01:14] Speaker B: to think of it as the iPhone.
[00:01:16] Speaker A: The everything came from one company.
[00:01:18] Speaker B: It all worked together, it was very integrated, it was more proprietary and hard to connect to. And there were very few APIs in the early days, there actually were none.
[00:01:26] Speaker A: But.
[00:01:27] Speaker B: But they opened it up over time. And then as the company grew, there were things that fell behind. They didn't have a learning platform, they didn't have a recruiting platform that was very good. They bought an engagement company, they bought a data analytics company. They bought a lot of things actually. And the story of it being one integrated homegrown system didn't really play anymore, but that was okay because they were still considered to be the architectural leader and did a great job of sales and marketing. And then of course, culturally the company is very interesting and in the sense that they really believe in taking care of employees, which really appeals to HR people. And so the sales teams and the service teams and the engineering teams have been really wonderful people to work with, which also makes it nice because these are hard systems to buy and implement. And there was a massive systems integrator market around this because the system was so integrated that you really had to plan when you implemented it. You couldn't just turn it on and then play around. You had to decide how you were going to set it up. And this is one of the things that later hurt Workdays that a, a lot of companies found that over their many, many years of using it as their job architectures and they did mergers and Acquisitions and other things that they had to restructure or what they used to call re implement workday multiple times and just because it's old is I think it was designed in 2006 and 7. You know, companies have had it a long time and companies change a lot over the years. They get flatter, they get more distributed or more centralized. And so lots of business rules and org structures and job titles and job
[00:03:00] Speaker A: levels and job families have to change.
[00:03:02] Speaker B: So all of has been a burden to workday customers. But you know, I don't know of another company that's been as successful as
[00:03:08] Speaker A: they have in hcm.
[00:03:10] Speaker B: And HCM is very complex. All of these applications are extremely complex.
[00:03:16] Speaker A: They are not as simple as you think. And they all have competing standalone vendors. So whatever you decide to do in workday, somebody will sell you something somewhere else that feels like it does the same thing. And what happened in the last 10 years is a lot of new ideas came to market and workday started to fall behind. I know workday pretty well. And for example, the learning platform was a little bit of a hack job in the early days. It picked up speed over time, but it was behind. The recruiting platform was somewhat behind. They didn't have a skills engine so they built one and they kind of over positioned it for a while and later it caught up. They didn't have a skills data import system, they didn't have a way to bring content into the learning system. There were a lot of little things that got left out because the system
[00:04:02] Speaker B: was so deeply architected in the beginning.
[00:04:04] Speaker A: They didn't have time to build all these things out. And the state of play in all of those markets kept changing. So anyway, but the company kept growing. And if you look at the company's revenues, it's almost a completely perfect growth chart growing at 15 to 20% per
[00:04:18] Speaker B: year, year after year after year.
[00:04:20] Speaker A: Very, very rigorously well run sales and financial operation. But the price to earnings ratio went down. The stock started to plateau. There were stories of people maybe not using it as much or maybe turning it off and switching to something else. Very rare. But occasionally the stock market kind of punished them along with the other companies as the AI world started. And it turned out that Anil Bashiri, the co founder, who's a very highly skilled leader and technologist a couple years ago turned over the CEO to Carl Eschenbach. And I, I don't know Carl very
[00:04:54] Speaker B: well, but I met him a couple
[00:04:55] Speaker A: times and he's really a sales executive. And the problem when with large software companies is when they're run by sales executives is sometimes they lose their edge
[00:05:03] Speaker B: to please Wall street.
[00:05:05] Speaker A: And that can be very dangerous. If you notice, Larry Ellison is really a software guy running Oracle. Satya Nadella is an engineering guy running Microsoft, Google, Meta, all of the really successful large technology companies are run by technologists. To turn over a software company to a sales executive sometimes works. It did in the case of ServiceNow, but sometimes not. So anyway, what happened for a reasons is more and more of Workday's messaging leadership team and product started to feel a little bit behind. And many of the Workday customers that we work with and we talked to, lots and lots of them were essentially forced to buy third party products because
[00:05:47] Speaker B: they weren't getting what they needed.
[00:05:48] Speaker A: And that makes a mess because the integration with Workday is problematic. It's a little bit hard to do. And Workday had a business model where they actually didn't let people integrate unless they were approved by Workday. And Carl tried to fix that and really open the thing up and start to do reselling and OEM and more partner activities. And I thought that was actually a really good idea. At least it, you know, opened up the market so that Workday could stay neutral. But anyway, many things fell behind and there were people involved that I think were the wrong people, at least from my perspective. So anyway, Aneel comes back from retirement or chairmanship a couple years ago and says I think I have to come back and do that Steve Jobs, Howard Schultz thing. And he decides thank you to Aneel, that he's going to really come back and try to reinvent the company for the next chapter, as he calls it. And thank you for doing that because things have really changed. New management team, the acquisition of Hired Score, then Paradox, then Sana. Two and a half. Almost $3 billion just on those three companies. New leaders from all those companies, new engineering teams from all those companies. Lots of AI savvy people now that were not there before. And now a complet new culture. A new head of product, Garrett, who came from Google or rather from SAP and Google. Lots of technologists who had not grown up in the workday world. And a whole new concept for what is Workday's role in the world of agents and AI. So let me talk about that and some of this is my own opinion and some of this is my own experience.
So because of the excitement around Claude and OpenAI and ChatGPT and Copilot and all of the possibilities of these generative AI systems, people are now convinced everybody that I know that we can build applications a hundred times faster than we could before. By the way, I lived this world in the client server days with powersoft, back in the days of Sybase, for those of you that are as old as me. So you can just sit in front of Claude and you can tell it
[00:07:51] Speaker B: what you want it to build and
[00:07:53] Speaker A: it'll build something for you and it'll run it and it'll work. Now, I'm not saying, sure, you can deploy it at an enterprise level, but you can build an app of your own very, very easily, certainly a website and a lot of functionality in that. And if you tell Claude or Codex ChatGPT to connect that app to another app, it will read or find the APIs and it will do a reasonably good job of connecting it. Then there's the issue of will it build the right database structures? Will it build the right business rules architecture? Maybe not yet, but it's getting closer and closer and closer. And so what's going on in the world of every technology company and IT department that I've run across is how fast can we build our own agents and get away from this slow, hard to change existing system from this big software company, Workday, Salesforce, SAP, whatever. And so the software industry, these big companies are in a panic of course, to reinvent their role in this new world of highly productive application development and generative AI. So you know, they're all taking slightly different approaches. Salesforce jumped in and claimed that they had all the agents. I don't use Salesforce anymore. I got tired of it. But I think the problem with that is that you're not going to buy all your agents from one vendor. And so if Salesforce thinks everybody's going to buy their agents for everything that they're doing in sales and customer service, they're mistaken. So that's one approach. A second approach is you add AI agents to your existing application, which everybody does, to make your existing application more functional and more interesting. Really. The acquisition of Sana and Paradox means that workday's done, that the recruiting system is being completely re engineered and reframed around Paradox learning same thing. Then SANA is going to be the layer of user experience and development tools. Also the third approach is you turn your company into a professional services company and you create forward deployed engineers. So OpenAI and Anthropic and Microsoft now have engineers that will go into your company and help you build agents so you don't have to figure out how to do it. And they Have. And by the way, so will ServiceNow. I ran into a company doing this with ServiceNow. So if you want to build an agent that does global onboarding or some other kind of very company specific thing, they'll come on site and they'll help you do it and they'll show you how to leverage the security tools they have and the infrastructure tools and so forth and they'll help you get it up and running. And that's good for them and for you because you get somebody to help you go through it and you and they get smarter and hopefully you get to learn how to use their products well. And nobody has everything and everybody's architectures are different. So you have options now and then the fourth option is design for the future. And I kind of think this is where workday is going and I think this is where I'd like to help workday go. What I find the most frustrating of the current situation of AI is not that there aren't enough tools, there are way too many and there's new ones being invented every day. But nobody that I can find yet has truly worked out what the agentic business system of the future would look like and how it would work. Now we have done that in HR and we're publishing it soon, called HR 2030. And it's a pretty detailed framework of how the agent world will evolve. And it's much, much different than you think. It's not automating a bunch of steps in the current process. That is a early stage tactical part of it for sure. But that's not where the big ROI is. The big ROI is redesigning these business processes around the users, not the company's internal processes. And I'm not going to stop harping on this because I think it's really important.
The way we're set up to do things inside of companies with job families and functional departments really goes back to the industrial age when we were either a railroad or a agriculture company, or a manufacturer or a car company where we had engineering and then we had marketing and then we had sales and then we had channels and we had finance, we had hr, we had it great. All those are important domains of expertise and functional operations inside of companies. But customers don't operate that way. Customers work across those domains. So customer service has to know everything about the products, everything about the pricing, everything about the support, everything about the market. And so to me, the opport and same thing with employees. Employees don't want to know who to call in HR for every little thing. That comes up, they just want an answer. So what AI does maybe more than anything else is it allows you to build what I would call horizontal applications that stitch together these functional specialties into journeys and solutions for customers or employees. And I think Amazon has done this extremely well. I think some companies have done this extremely well. But they have to do it despite the erp. The ERP system didn't really do that very well. In the ERP system you could run a report or you could build an application on top of it that did that. But the ERP system pretty much automated the functional areas independently and then created all sorts of reports and analytics to analyze the cross functional areas of the business. So I guess my point is that so far no one has clearly articulated what an agentic business system would look like. In other words, how would you architect the agents to pull all of these functional expertise areas out of the functional silos that they exist in today to operate your company more effectively? And it's probably not going to be one architecture. I think every company is going to do it different. I mean, one of the banks I went to this week is very focused on their high net worth individual offering. Well, if you just sat back and you were a bank and you said, what are all the things we want to do for a high net worth individual? You could put it on a whiteboard in a day and you could sort of pull services from different parts of the bank together and say, okay, let's build an agent that pulls all those services together and personalizes them for each individual customer. I'm sure banks are working on that. That's not the same as buying an agent from a vendor that automates one part of finance or HR or recruiting or whatever. That's a completely different idea. And then the question comes up, if you built this high net worth service and sales and investment agent, which would have a lot of things in it, who's in charge of it? How does it know what data to get? Who has access to it, how do we know if it's making mistakes, how do we keep it up to date, et cetera. Those are things that we've spent 50 years, 40 years working out in it, going back to the mainframe days when it was all centralized to client server and beyond. But we haven't really done that with agents because agents are self deterministic learning things. So we got to give them guide rails and rules and access permissions and stuff so they don't accidentally disclose information or make mistakes or go outside the boundaries of the Rules we want to create in the company. We're working on a really interesting project with Microsoft right now to create rubrics or rule frameworks to maintain the integrity of these agents. And those can't be coded into each agent. It's not going to scale that way. So we need ways of organizing these things. And one of the things that occurs to me, and I'm not a software architect, but I'm an engineer, is, you know, maybe we need types of agents. We need agents that monitor things, agents that set rules, agents that get control access to data, agents that control access to decisions, and functional agents that deal with user needs and user journeys. And maybe there are different categories of agents as opposed to saying, let's stick it all in some framework on top of everything else. So anyway, so what Workday did is they said, okay, we see this problem and we're going to deal with it and we're going to take it head on. And so they've built a set of tools for it and for developers and users to register agents, keep track of who has access to what. And a passport, I think the word passport's a great idea, where you could get stamps to get access to different parts of the company systems and data, or have them revoked if your agent behaves poorly. So what they're basically saying is that Workday is not only a system of record anymore, it is now a system of agents or a system of managing agents. And implicit in that vision is the fact that they have a development environment for you to build agents. And SANA is pretty cool. I mean, the new stuff coming out is going to be very attractive. There will be alternatives. There's lots and lots of guys building, you know, development tools. But you're going to like it, especially if you're a workday shop and you like Workday. And the question, of course, is, if you build an agent in Workday using sana, how interoperable will it be with non Workday stuff? I think they plan on making that easy because there's a version of Sauna called Sauna Enterprise that connects to other systems. But there'll be a lot of head scratching, I guess, is the word about whether we use SANA for this or that, and, you know, should we use ServiceNow for this or that, or do we use Copilot for this or that, do we use Claude for this or that and so forth. It's a pretty, pretty hot space and there's lots of alternatives. But I think Workday's story is going in the right direction. And they're really going to push this now. The second part of the whole reinvention is the structure and culture of the company itself. I know Workday pretty well. I mean, I've never worked there, but I've met a lot of people that work there and been in and out of the place many, many times. It's a very kind and gentle, hardworking, passionate, creative place. I think it lost some of its invention in the early days. It was very inventive and very creative. And it was. There were a lot of technology visionaries there. I think some of that is coming back now. So two things that are very significant. Adam Godson, who's the CEO of Paradox, is now running Paradox and the whole recruiting business and technology stack for workday. So that's a huge area of opportunity and it's a model for workday to build general managers that own different parts of the offering. And in parallel, Joel Hellmark, who's the founder of sana, now owns and runs the learning business, the entire learning business, apparently, including the old one and the roadmap for AI and all of the sauna development tools. And these are two young AI savvy executives that have both run companies that reached a reasonable size on their own, worked with lots of customers, understand the product, technology, marketing, sales support issues that are needed. And that's a big difference from having this big product group in workday that tries to stitch together a roadmap of all the different things going on. And if you go in, if you're a workday partner and you go into the workday support portal and you look at all the PowerPoints and stuff in there, you see how hard it is to keep the pieces together when they're all in one product organization. So I think this idea of having general managers and smart younger and I don't think it's an age thing, it's more of an evolution thing. People that really know what's going on in the world of AI is really going to help Workday innovate in a big hurry now. I mean, I think there's one other thing that sort of comes to my mind as I think about this. Is this a proactive, forward reaching strategy or is this a defensive strategy? Wall street is going to pick this up in a hurry. And I think the way Wall street works today is Wall street rewards vision and forward looking momentum, not protection and revenue through ownership of a customer. My example of that is Tesla. You know, I'm not a big Tesla fan because I don't trust a lot of things the company says. But Elon has done a pretty spectacular job of painting lots and lots of visions that people really get excited about and they buy the stock and it's given him market value and capital and investment dollars and customers that have really, really been spectacularly powerful for that company and his other companies and the companies that have to create a strong vision of the future. The stock market's really punished them. I mean, I think Salesforce has tried but not succeeded. I don't think Workday succeeded at this yet. You know, obviously Google and Microsoft are pretty good at it and they're well along on lots of new things. But for the traditional 20 year old or more software companies, it's been really hard. Adobe, DocuSign, all of them have struggled to clearly articulate where they're trying to go. And I think it's because it's not clear where they, it isn't clear where they want to go and they have to figure that out. So my hope, and I guess the thing try to help work they do is to get further clear and figure out with customers, for customers, what is this agent business system, agentic business system of the future going to look like now? My experience in this domain goes way back to the 1980s. And so let me just tell you my little story here because it might be interesting. When I got into computing in the 1980s, I went to work for IBM. And it was back when IBM was a mainframe company, before the PC had been launched. And the thing that amazed me about the world back then is that every major company we talked to was relying on IBM to teach them how to run their data processing department. It was called data processing at the time, wasn't called it. And IBM wrote books on this. There were, they were called red books. There were technical documents, how to organize things, how to manage data, how to manage transactions, how to manage scale, how to manage user experiences. IBM wrote the very first paper on the economic value of sub second response times. And this is in the 1970s. So they were thinking about user experience a long time ago. And because IBM was so big and really had a sort of, I don't know, paternalistic view of the world, at least it seemed like that at the time. People just did it. They just adopted IBM's practices because everybody else did. And then the world fell apart. We had open systems, we had Unix, we had open databases, the Internet, tcpip. And so all of the proprietary structures that IBM was selling slowly went away.
And there wasn't anybody teaching you how to do this. So everybody winged it and lots of huge companies were born, including consulting firms helping organizations figure out how to manage and develop their technology. Well, that's exactly where we are. Again, I don't know yet of anyone who's really described how the business architecture of an agent system would work. I'm sure people are thinking about it and I hope somebody works on this and it might be Anthropic, it might be Microsoft, it might be ServiceNow, it might be Workday. I particularly hope Workday works on it and I'm going to try to work with them because HR is such a big part of this. I don't know finance the way I know hr, but I think we have to give people a vision so they know where to focus their attention. Without that kind of a vision, the market is very confusing because every company I go to has a vendor stack already and then a bunch of favorite vendors that are selling them new functional things that seem like magic and you know, it ends up becoming a lot of third party products that are somehow connected together in some form of architecture. IT people are all over this. There's lots of things going on to try to build frameworks to connect it. And I think Workday, because they're very savvy IT centric software company can really help sort this out. So I'm very bullish on Workday about all this. I also think from a pure financial standpoint, the products from Paradox, Sana are high margin, high revenue, groundbreaking products.
So those of you that haven't seen the new Workday stuff from Paradox or Sana, you should see it. You're going to want to buy it because it's really good, they're really good products. So there's a revenue stream here in those two areas that's huge. The recruiting industry is a good 200 billion, in the learning industry is a good 400 billion in size. There's lots of money to be made here. So I think Workday is going to do well with that. And the culture of the company is still very strong. I mean, it's a very, very well run culturally company, even though it may have lost a little bit of its innovative edge. So there's going to be a lot of good things that'll happen. So I'm pretty bullish on what's going on over there. And we'll keep you informed. And we're working obviously very closely with Workday because we're a partner now because we built Galileo on top of Sana and so we know know as much about Sana as many people, most of you. And we're, in a sense, integrated into workday, out of the box. Okay, that's it for now. See you guys later. Bye.
