Episode Transcript
[00:00:00] Good morning, everybody. Today I want to talk about the job market. I want to talk about HR as a profession, and then a little bit about my experience at the TRANSFORM Conference in Las Vegas this week. First of all, the job market. There's a lot of articles and stories and news about layoffs and restructuring and threatening jobs, eliminating jobs, decimating jobs, destroying jobs. And I took some time over the last few days and I wrote a fairly detailed article that tries to debunk all of this. And it was interesting. While I was writing it and reading a lot of stories about this topic, I ran across a podcast from Jensen Huang, the CEO of Nvidia, and he actually agrees with me that AI is not destroying jobs, it's changing them. And what it's really doing, as you know, is whatever you you do for work or for a profession, it's automating a whole bunch of steps. And when you automate steps, two things happen. First of all, you as an individual decide to do more things or other things with your time, because the things that used to take a lot of time happen quickly. So if you're an analyst and you used to spend a day collecting information on companies and then spent another day analyzing it and you just saved a day, you now can do twice as much analysis, or maybe one and a half times as much analysis. If you're a salesperson and used to be going through tons and tons of leads to figure out who to call back, and calling a lot of people that didn't want to hear from you, the AI gives you a qualified list. You just saved the day or two of going through leads, and you can call the people that are more qualified more quickly. That goes through. That's true for software engineers, that's for managers, for nurses, for everybody. And as you'll read about in the article, I talked explicitly about software engineers and then about X ray technicians and nurses. What has happened in the last two years since AI has started to pick up speed is the number of jobs have gone up, not down. And the reason for that is that as pieces of your profession become more automated, your profession adds scale and you can do more things and more people consume. In the case of software engineers, as you'll read it, the full stack engineer of today has different skills than the full stack engineer of 10 years ago. But that's not to say the engineer is not needed. We still need people to build things and engineer things and design things and build infrastructure stacks and test things and all that. It's just that the AI makes it happen faster. So the story about eliminating, destroying jobs isn't really true. Now, of course, as you all know, it's no fun to change professions or careers thanks to technology.
[00:02:55] So those of you that work in highly routine roles, you may find yourself doing something different. For example, if you're a driver and you're driving for Uber and there's a bunch of autonomous cars entering the Uber fleet, maybe you won't be a driver for the rest of your life. Maybe you do something else. Jensen made a funny comment that I think has a lot of truth to it, that maybe what'll happen is the autonomous car will have a chauffeur in the car who will help you with your bags and help you do other things, and the car will drive itself, or the autonomous cars will be owned by a fleet company, not you or the driver. And the fleet company will have a staff of people supporting the enormous number of people using autonomous cars, by the way, which will be greater, because the autonomous cars are more efficient at driving and can take less traffic because they know how to stay away from each other. And now there will be jobs in the fleet company to support these cars that may not be the same as driving them, but they may be similar. So every one of these situations where something that used to be manual became automated improved the scale of that particular role or business, creating new opportunities for people. And anyway, I spent a lot of time writing about this, and this is the super worker, super job, super company effect. And for those of you that follow what we do and what I do, I'm working on a book on this that will come out in the fall, fall. And we'll be giving most. I think we'll be giving you all a preview of the book if you come to our conference in June. Okay, so that's number one. Number two, there were a lot of interesting announcements from product companies this week, and there will be more to come next week. And the theme of all of these HR tools is not necessarily more sophisticated applications, but breaking down the infrastructure barriers between systems. And let me just talk about that for a minute. Back in the old days of mainframe computers, for those of you who even know what a mainframe is, there were very few software tools to connect systems together. So if you wanted to get a software tool that did recruiting stuff to talk to a software tool that did pay stuff or skills stuff or whatever it may be, you basically built them together in the computer, in the mainframe. And then over the 20 or 30 years since then, we ended up with PCs, client, server, cloud, systems and APIs. An API is an application programming interface, and it's basically a set of rules where a software engineer or company says, here's what our software does, here's what the database looks like, here's how to get access to all of the things in it. And so if you want to interact with our tool or system, you can read this API and program in a set of interfaces to get to it. And that took time and engineering. And it was problematic because the APIs needed to be tested by the company that wrote them and the company that built the interfaces between them. And it created a lot of interconnectivity opportunities, but it was slow.
[00:06:05] And, you know, 80% or something of the applications in big companies are not fully integrated with other applications. And IT departments don't want to spend their entire lives integrating the 400 applications the company buys just because they want everything to work together. They would like the vendors to do this for them. So an industry of middleware companies was formed. Mulesoft is one that was acquired by Salesforce. And there's a whole bunch of them that build software that connects software to software. Most of you probably don't even know about this and don't care about it, but it's kind of an interesting part of the market. Well, the actual activity of doing that can be automated by AI. In other words, AI can read the documentation about Workday or SAP or Oracle or Telayo or whatever it is you built, assuming the engineers make the interfaces available, and it can decompose them, understand how they work in workflows, business rules, and security rules, and it can build integrations between systems. So all of a sudden, this huge market with billions of dollars spent on integration software looks like it could start to disappear. And that is a very significant thing for those of us that work in business, because this has been a major obstacle to the employee experience, the candidate experience, the experience of a recruiter, the experience of virtually anybody in the corporate world. Next week, roughly a week from today, there will be an announcement about this from one of the vendors that you may or may not know. That is very interesting and very significant, and this will change the market. The company that has been, you know, much ahead of this is ServiceNow, who's basically built applications that sit on top of other applications to make your life easier, because the vendors you're buying from aren't building everything themselves. And then there's the issue that some companies don't want you to connect to their systems deliberately. Oracle's a little bit this way. Although they're not intentionally this way and make it even harder for you to connect things together. I want to kind of give you a heads up about that and I'll talk a lot more about it next week. Third thing I want to talk about is a really interesting company we've been working with for a couple of years called Findom. Findom is in the recruiting market and they've really done a groundbreaking thing. And if I spend five minutes, you'll understand what it is. So when you buy software or you go to LinkedIn, you use the recruiting, the hiring assistant, or you put ads on. Indeed. Or you post jobs on your website, or you hire a company to search for candidates for you, or a headhunter or whatever it may be, your recruiting function is sourcing candidates. You know, you may be advertising and letting the candidates apply. But if it's a rarefied, somewhat specialized job, you're going to want to go out, look for this person. And when you look for them, you use talent intelligence software like Eightfold or Seekout or others. What they do is they have amassed billions or at least hundreds of millions of profiles which are, by the way, available for sale. All of the data about you that you have put on the Internet is available for sale. A lot of companies have scraped it and collected and cleaned it up. And they build a tool on top of that, using AI to find people. And in that tool you say, I want somebody with 10 years of experience who has these certifications, has these skills, that's had these job titles, that speaks these languages, that has this college degree that lives in this city or this city, speaks these, you know, understands these technologies, whatever. And the system the software tries to figure out from these billion profiles, by the way, a lot of people aren't in those profiles. Certainly people in certain countries aren't there, like in China or maybe Japan. And, you know, certain people in operating roles, like frontline workers would probably not be in those databases, but they're getting there. And the software tries to figure out who the people are that best meet your criteria. And it uses the mostly data like LinkedIn, which is historic data about your career. So it's trying to guess based on the job titles you've had, the companies you've worked for, and a little bit about what you've written about your bio, what you know how to do. It doesn't really know that much about your activities, but, but it knows sort of your. In the imprint of what your career looks like. Like, if you looked at my career, which is a little Bit, you know, unique. You would probably figure out I know a lot about technology because I worked for a IBM and Sybase. I know a lot about marketing because I've been the VP of marketing several times. I know a lot about sales because I had a business development job. And then you'd say, well this guy seems to be a little bit all over the place. Oh, he's also been a CEO, so maybe he's a general manager type also. You may not know that I'm an analyst and that basically that's my main thing. But you know, you, you could figure out parts of it. So what happens is when you have a unique job like the general manager of a media business in Japan, or a whole bunch of salespeople who are operating in Eastern Europe, or for a product that's new in a certain domain, or a bunch of biological scientists that are working on a new MRNA solution in a bio company or whatever, or marketing specialist who understands a certain market, your software is guessing that these people, it gives you this list of people have these qualifications and because the software is so sophisticated you get a ton of people and now you've got to go through this list and you've got to sort through them and qualify them. Now there are better and better tools for doing this. And the pre assessment market is becoming AI enabled too. So vendors like shl, Machi, People, Predictive Index, Predictive Index and others have built assessments so that if you can get these people to take the assessment, if they're willing to apply, you could learn more about them. But that's hard too because people that are really good at their jobs aren't applying for other jobs, they're just doing them. So this search problem is it is what it is. Well, if you look at what's been going on in the LLM market, which is the big frontier vendors, Google, OpenAI, Anthropic, et cetera, they've already gone way beyond that. And what they do is human labeling of data. If you looked at a whole bunch of resumes and you're a recruiter, you're going to notice things that the software doesn't notice. You're going to notice people that have worked for fast growing companies, you're going to notice people that haven't worked for fast growing companies, you're going to notice people that have worked for global companies in certain parts of the world. You're going to notice people that haven't worked for global companies. You're going to notice people that have jumped around a lot or you're going to notice people that have progressed into promotions in a single function. You're going to notice a lot of things that the search software may not understand.
[00:13:16] What if the humans labeled the data and what if you told the humans and let's assume these are humans, but they're not really humans. They could be humans. You said there's a rule. We're going to create a rule that says anybody that's worked for a company that has doubled in size in less than six years, we're going to tag them as fast growing, company experienced, or anybody that's moved up one or more levels in their jobs twice within a five year period is a an ambitious person or somebody who's rotated around the world into this country, in this country, has Middle east experience, or has Eastern European experience, or has Asian experience, whatever you want to call it. That labeling process is a second order way to analyze and search and categorize data. And we do this a lot in our research. I mean, this is really what our research is about. When we do maturity models, it's not a simple statistic. We look at a lot of dimensions of data to figure out what level one, level two, level three, level four are. So what Find Them did and this, and this is what the LLMs do. And they do this to make sure that they're not getting junky data into their platforms. And they hire people to do some of this labeling. Like they hire people to label images, they hire people to look for porn or things they don't want in their system. And then they built a whole variety of systems to automate the work that the people do so that the human labeling is more structured and scalable than one person going through one record at a time anyway. So Findom took all of this, these ideas and technologies and applied them to hr. And so what Findom does is you take this massive list of people. By the way, all the people in your company are also a source of data. Because if you're doing an internal analysis of who would be best for a job and who would be best for a senior role, who would be best for a project, who needs promotion, who's behind, who's likely to leave, you want this labeling too, because you won't be able to really know from just looking at their name and their title because you may not even have their job history in your HR system what characteristics are correct or none about these people. So this labeling initiative or idea or technology and solution from Findom is really valuable. And we've been spending a lot of time with them. Yesterday I was at dinner with a bunch of Findom customers, and one of them who will we're going to have an interview with her is a headhunter from a large global headhunting firm. And she told me she's looking for senior executives in several countries in Asia to run this entertainment business and to run, operationalize all the different functional areas of this entertainment company. They're doing a big search of all sorts of roles. And she said none of the traditional tools would even help us because they didn't even have data on these particular countries or these particular roles. But the Feynum data was, as she called it, a game changer. Now, I'm not here just to promote Findum. Findum's a very interesting company. I think a lot of you are going to want to talk to them. But I'm here also to explain to you this idea that the data that you're collecting for your talent management system, your talent intelligence system, your skills inference system, whatever you may be calling it, if it isn't labeled, it could be quite misleading or less useful than you think. For example, in the recruiting domain, one domain of labeled data is people that worked in the military. If you worked in the military and I didn't, but those of you that did, there's all sorts of job titles and roles and job descriptions and certifications and descriptions of who you are that are very well known in the military, but not well known in the business community. So what companies like Findom has done and Eightfold has done this also is they've labeled their data sets for military candidates. So now they look like, from the data standpoint, people that have worked in business jobs. And you can see somebody who's been a senior executive in a sense on a ship or in a fleet or in some other role in the military, or an engineer or a scientist or somebody who has a lot of people skills. So this is a really interesting area and it's going to be evolving a lot over time because more and more of the systems we use for HR are analyzing the data on the people in our companies, not just the people outside our companies.
[00:17:46] Okay, final thing, I want to sort of say a little thank you to everybody who's listening to this. A lot of, you know, I've been doing this a long time. I love this work. I love this space. I love all of you. I love the people that I work with in our company. And yesterday when I was at this dinner. No, these are people that I don't really know many of These people work for medium sized companies, some large companies. Three people came up to me and just said, I want to thank you for what you've done for the last 20 or 30 years. I've been following you, I've been reading you. You helped me tremendously in my career. You helped me understand things, you helped me get promoted, you helped me find new opportunities. Thank you, thank you, thank you. And I just want to say thank you to all of you listening. I really do this because I love it. I also do it because of the wonderful feedback I get from you and the impact that we can have on you, your companies and the world. So just being at this particular conference and having so many people come up to me and tell me how much they appreciate the work that we do in our business and all of us really sort of made my day. So I want to say thank you to those of you that are listening and one small plug or explanation. The reason we built Galileo wasn't because we really wanted to be in the AI technology business, although we kind of are. It was really a different reason. We are. We have tried for years to find ways to scale what we do to more people at a lower cost. And, you know, you can call us on the phone and we can do a workshop and we can do a webinar and you know, we can have conference calls, but there's only so many hours in the day and the week. We're only, you know, we're a finite number of people. I'm only one person. And so what Galileo is, is it's the digital manifestation in a very, very powerful format of everything we have ever learned. And we're going to keep doing that. So I really want to encourage you to get Galileo, and I know you do have to buy it and I know you have a lot of other tools and you maybe are not sure if it's going to do what you want it to do. I just want to assure you that it is spectacular way to, to dive in and understand all of the things I'm writing about or we're podcasting about, or we're talking about in speeches, or we're talking about in meetings. Because not only is all of our research in there, but benchmark data, third party data, skills data, job data, economic data, company examples. The way we've tuned it means that you can ask it a simple question, you can ask it to build you a development plan, you can ask it to compare options, you can get it to advise you, and you can turn it, put it into what we call agentic mode, and it will ask you questions to diagnose what you're working on.
[00:20:35] So thank you for all of the feedback and wonderful positive reinforcement that I got this week. And that's it for now. Bye for this week and talk to you guys more over the weekend. Thanks.
