Episode Transcript
[00:00:00] All right, you guys, today I want to spend 20 minutes talking about the candidate agent market and what this means in terms of AI in general and recruiting. And highlight Eightfold who just announced their candidate agent. And I won't talk too much about Eightfold, but I will talk a little bit. So in the process of recruiting or selecting people internally for any job, there's a whole conversation with the candidate about what the job is, when it is the hours, the location, the certification requirements, the skills, the expectations, the pay, the rewards, the benefits, the flexibility, et cetera. Do you need to be licensed? What is my career path in this job, how do I grow in this job, et cetera. And in the high volume world of recruiting, when there's lots of turnover, these are questions that are fairly mechanical, that need to be asked as qualification questions. But the candidates have questions too about the company, the job, the location, the work that they want to ask. And if we go through a recruiter to do all that, it takes a lot of time and it has to be scheduled and it delays the process. So a candidate agent is a really big market. I think Paradox pioneered this, was the most successful. There was a big market for candidate chatbots maybe 10 years ago that kind of fizzled because we didn't have good AI. I think Paradox proved that you could build something pretty spectacular out of natural language processing. And that turned into Paradox now part of workday. So Eightfold has built one of these now, and Eightfold has lots and lots of intelligence in its core product strategy and also in its data systems for talent intelligence that can be used for this application. Now, in the HR 2030 world, where we think about the future of HR in the over the next three, four, five years, the candidate agent is potentially also an internal candidate agent. So let me just take a minute and talk about that. The external candidate and the internal candidate have a large overlap, but they are not at all the same. The external candidate has a lot of questions about the company.
[00:02:12] And we don't necessarily know the external candidate well unless they volunteer to give us access to more information.
[00:02:19] The internal candidate, we know a lot about them. We have their whole job history in the company, their pay, their skills, their certifications, their performance reviews and so forth. But we may not know their goals, their aspirations, why they are trying to find a new job, and what this new opportunity would mean for them. So the behavior of the agent for an external candidate and an internal candidate is similar, but not identical at all. And the data sources for the external candidate agent and the internal candidate agent are also different. And then there's the question of what do you want the agent to do as its outcome? Does it make decisions or does it just collect information and produce that information for the screening process or face to face interview process that comes later? Do we want the candidate agent to schedule the interview or not? Do we want the candidate agent to potentially block out people who are not qualified based on questions that are asked? Those are very company and job specific things also. So this agent has a huge potential use case in many of these important areas. In high volume hiring, it's very often used for some amount of screening because if the hours are, you know, 3pm to 9pm and the candidate has kids at home and they can't leave their home during those hours, that's important to determine early before we ask all sorts of questions about their background and experience. In a more white collar or higher end role job, it's more likely to be a career development coach, not just a candidate Q and A agent where the candidate is asking the company, in a sense the agent, well, what are my opportunities in this job? What will happen to my pay over time? How much development will I get? What's the first year experience like? You could imagine a young aspiring high potential person would actually want the candidate agent to be a career coach. And MasterCard does that. L' Oreal does that with their recruiters. And you know, this is why sometimes humans are better than agents. So we have to decide when you buy one of these things or build one of these things in this wide range of use cases internally and externally, where do you want it to focus? I think if you look at the most scalable implementations of candidate agents today, they are in high volume hiring where the questions are pretty predictable. And we're not trying to turn the agent into a development coach or a development expert. But I think it will only be a short period of time before these agents get turned into onto white collar jobs or internal candidates and they're going to have to have that capability as well.
[00:05:06] Now in the case of Eightfold, let me just mention them, there are lots of these out there successfactors, smart recruiters has one. And I would guess that almost every ATS vendor or significant talent vendor has some form of an agent for candidates to interact with. In the case of Eightfold, Eightfold also has an AI interviewing agent which is very different. The AI interviewing agent, which is another category of agent, is actually a much stricter, more IO psychology trained tool that forces a candidate to sit still, takes pictures of them to make sure that they're not gaming the system and does other things to try to verify if the person is real. It may ask them technical questions, it may ask them, it may give them quizzes, it may ask them developmental questions. It may ask them questions based on your own IO psychology assessment of what are the skills and capabilities of this job.
[00:06:03] Machi people has this kind of technology, as do Hirevue a lot of others. So the chances are that the candidate agent will take you to some level of interest where somebody wants to apply, and then they apply and, and then they go through the screening and interviewing agent. So the way the world works today, that's most likely going to be two completely different products, two completely different AI agent solutions. But I could certainly see them connecting together, you know, pretty quickly because the interviewing agent or the AI interviewer, which may have been trained on job skills and job qualifications and patterns of high performers in a given job and so forth, could get a lot of information from the candidate agent to inform its assessment before it starts the interview. It's sort of like if you had a human interviewer and you were talking to the recruiter about the job and asking them all sorts of questions about the qualifications and the hours and the pay and so forth, and then the recruiter started asking you a bunch of questions, you wouldn't expect the recruiter to forget everything that you just told them.
[00:07:10] So, you know, I could be pretty easily convinced that over the next year or two, and this is part of HR 2030, these two agents are going to be much more tightly linked together. Now, there's a few architectural issues here too. When you, if you think about Eightfold or Seekout or some of the big talent intelligence companies, which, by the way, most of the sourcing and recruiting companies are talent intelligence companies, also the agent could tap into data about the candidate that the candidate doesn't know it has.
[00:07:41] So I know in this, these days of worrying about AI, this may not be the greatest thing to do, but I think it would be pretty okay. And I think most legal departments would be okay that if the candidate opts in to allow the interviewing system or the recruiting system to access information from other sources, these companies have a lot of data about a lot of people. I mean, Eightfold, for example, is trained on 1.6 billion profiles. Now, they're anonymized in the system, but you can get to them if you need to. So I could certainly see a direction where the candidate agent says, are you the Josh Burson that lives in such and such a street At Oakland, California? Yes. Oh, well, we see that you've done this, this, this and this and that. You've applied for this job and this job.
[00:08:34] I mean, that may seem a little creepy, but we're very close to that kind of a system too. And I think that's a sort of solution that you want to work on with your vendor and with your legal department. That could be very, very powerful. The other thing, of course, where their data is involved is if this person worked for the company in the past and quit or left for some reason and is coming back, or if this person applied before and was interviewed in the past, usually called Silver Medal, they may that data would be useful to the candidate agent. So there's a potential for candidate agent to be pretty intelligent about, you know, this particular person's interests. Now, the second part of this whole space, which comes out in 2030 a lot when you look at the 2030 architecture, is the relatedness of this agent to the development part of work.
[00:09:31] So think about maybe let's just talk about an external candidate who applies for a job. They ask all sorts of questions, they get all sorts of great information. They decide it's a good fit, do the formal application process.
[00:09:44] They go through the interviewing process, whether it be AI or video interviewing or face to face, or maybe they even get in a plane and go meet with people and then they have a start date and they have a, maybe a couple of forms to fill out and maybe a few books to read and the manager is ready and they show up the first day of work. The manager knows nothing about all the information that was just gathered about this person. They may know what they learned during the interview process, but all of the other conversations they may not know unless they spend a lot of time with the recruiter. That's the traditional way of doing this. But the AI has all that information.
[00:10:20] So why wouldn't this AI super agent, we would now call this more of a super agent, take the information that it gathered from the candidate agent experience, from the interview process, or interview agent experience, and map it against the skills, capabilities, experiences and backgrounds of the high performers in that job, which by the way are over in the performance management database, and say based on what we know the high performing people in this job do, there's a few gaps this person's going to have. We have a standard onboarding program to understand our systems and our processes and our culture and our customers. But we also have a custom bunch of things that we've identified as opportunities or gaps for you and why don't we put that together into an onboarding program custom to this person? I know for a fact that you can do that today. You can do that today with Galileo Learn. Galileo Learn actually knows as you use it what you're trying to do and what level of skill you have in a whole bunch of different areas because of your behavior in that system alone. But imagine how much information you get when you enter that part of your job process from the recruiting process. And it's always struck me for many, many years that recruiting or talent acquisition was like on one side of HR&L and D was way on the other side. There was like this dumbbell effect where the two centers of excellence were very, very unrelated. Well, actually they're very close.
[00:11:58] There's a, there's a good argument that they might want to be under the same leadership because the talent acquisition process, internal and external, really pulls out a lot of experiential skill, background, motivation and tacit knowledge as well as technical knowledge from an individual which should be used in their development plan to succeed in a role. I know in our company where I know the company very, very well. For example, we have a way of doing what we do in research and we have a way of doing what we do with clients. We're a very high touch, consultative company. It's not that hard during an interview for those of us that have been doing this for a while, to look for blind spots in somebody's background and identify things that are likely going to be either blockers or what we used to call pushers, in other words, benefits in this person's role in our company. I happen to know this because I've been here a long time and done this a long time. The AI would know this with real data by looking at the characteristics of the high performers. So I think the way this is going to work in HR 2030 is there's going to be a screening or candidate agent that answers a bunch of questions. There's going to be an interviewing agent that does all the interviewing qualification stuff and then there's going to be a development agent that takes that information from the non employee. By the way, the employee obviously has much more information coming through the candidate process and maps it against the known skills and capabilities of the high performers that it knows about through the performance management agent. And you can see how this works is this little chain of four agents with a few others added in for fun is basically what HR 2030 is all about. By the way, in the HR 2030 agent framework or blueprint. There are more than 130 agents acting together like this because there's a few other things I left out here, but this is the big, big benefit of AI. I think most people will look at candidate agents as a way to save time, reduce the number of scheduling complexities and reduce the number of interviewers and give the interviewers and recruiters time to do more strategic things and to improve the experience for a candidate. Because if the candidate's going to the website in the middle of the night, there's nobody to talk to. They're just going to go away. But if the agent is there, they can talk to the agent and learn more about the company and the job. So there's all sorts of benefits as is. But as this agent network is built out, and I'm quite sure this is going to happen based on the vendors that I know, then the information that we capture in the internal or external recruiting process should and will be used for development, planning and onboarding. And, you know, I don't think of onboarding as the first day or the first week or the first month. I think of it as the first year. Every job I've ever had, when I changed companies or changed roles, I mean, it did take six months or a year to really feel like I knew what I was doing. And that's not because the job was particularly complicated. It was. A lot of the roles I had were very different and new. I had, I had a lot of. I had a career where I jumped around a lot. But the company is different. The people you need to meet, the business processes, the systems, the customers, the way the company goes to market, all sorts of things take time to learn. And I think a great development agent, you know, a little bit like Sana or Galileo or others, would be able to take the data from the candidate agents and the interviewing agents and the data from the performance management agent, which is looking at high performers, and pretty easily put together a development plan with real content. And that intermediate agent could also go to the manager and say, we have a gap in these capabilities in the new pool of people that are coming. Or that agent, as a super agent could go to the manager and say, based on the success rate of these high performers versus these moderate or lower performing people, we should change our qualifications and our interviewing process because the people that are succeeding had these kinds of successful outcomes from the interviewing and candidate experience versus those, this is going to turn into the equivalent of an autonomous car. And I've talked about this a lot, but let me explain it because we're in this very simple domain here in an autonomous car.
[00:16:38] Years and years and years of training build a data set that the AI knows that when a mother and child walk across the street, it's going to look roughly like this, it's going to move roughly like this and you need to slow down and stop.
[00:16:55] Right? You know, and there's a lot of training and image recognition that teaches the system about that one scenario. And there's hundreds and hundreds of scenarios when you're driving, obviously. Well, if you think about the work process in every company that I've ever been in, it's pretty much the same. You hire somebody, you select them and go through all that sourcing and assessment process. You onboard them, they start working, they have some feedback from their manager and from other people, they develop a reputation, they get a performance review, they might get, a couple years later, they get promoted, they might have a talent review, there's some succession reviews, they might have some coaching, executive coaching. Eventually they move into a higher level role or they, well, all of that developmental experience that happens if it were captured by AI, the AI could look at any given company, by the way, every company is different and say, you know, the people that have really succeeded in this role, maybe it's an oil and gas engineer or a salesperson or a repair person or a consultant, has been very good at these things and has done these things and the data isn't interpretive, it's actually just statistical data. And therefore we want to take other people that don't have this pattern of experience and either develop them or offer them experiences like our high performers. That's actually what AI is going to do. And that's an autonomous system because eventually with enough data in there, if you're say an investment banker and you're trying to build first five year career expertise in investment banking, you know, that's a very unique particular role. But take that, take that just as a generic example. And you look at the guys that are making, you know, bringing in hundreds and hundreds of million dollars of deals 3, 4, 5 years from now and go back and look at what they did and where they came from and what their experiences are and their performance management and feedback and so forth. You know, by the way, this is what some of the investment banks have tried to do with custom systems. I actually talked to a couple of them about this. They didn't have enough AI to do it well, but now you can. Then the system could autonomously go all the way back to the sourcing and recruiting process and say this particular candidate has a very low likelihood of being the high performer that we want, it wouldn't say no. It wouldn't necessarily block them. Could. But it would give you, as a manager, as a leader or an executive or an HR analyst, insights into how to score people. And that's where this is going. And I think the ROI of that is 100 times higher than scheduling interviews faster now, right now, scheduling interviews faster and giving people a better candidate experience is great. That's good for your brand, it's good for your recruiting, the quality of recruiting, the speed of recruiting, your experience in the market. You know, a lot of customers, a lot of companies that have consumer product businesses want the job seeking experience to be good because it reflects on their company as a consumer company. If you love Starbucks coffee, but you apply for a job and get treated like dirt, you're not going to go back and buy more coffee as a customer. McDonald's, all of them, Chipotle. So the customer experience is impacted and important and it's related to the job seeking experience. But this other scenario where we have a more integrated end to end process is going to be really big. Okay, so look at the announcement from Eightfold that just came out. Take a look at Paradox. Take a look at Mackie people. Take a look at Seekout. Take a look at Find Them. Take a look at Raidency. There's some amazing new AI capable talent acquisition systems. SAP is in the middle of upgrading their whole talent acquisition suite through the acquisition of smart recruiters. And it's really pretty impressive what you can do now. And just in the case, one more note on Eightfold. Eightfold is a company we've worked with since they were founded. They were particularly interesting because they've also announced an open version of their platform called Talent Forge.
[00:20:57] So you can take all this technology for sourcing and interviewing and also for candidate and tweak it and add things to it to make it yours because you know these agents are yours. This isn't something you buy off the shelf like an assessment and you just use it. You're going to want this agent to speak to your language, your culture, your issues, your jobs, your roles, your management needs. And so you're going to want these platforms to be highly configurable and in some sense programmable like development tools themselves. And that's what Talent Forge is about. That Eightfold introduced at their conference a couple months ago. So I hope this is educational. I'll be writing an article on this. Stay tuned for that and congratulations to Eightfold for getting this out the door. For more information on HR 2030 and seeing how the big picture comes together, take a Look at the HR30 page on our website, download the introduction, and then if you become a corporate client of ours, we'll take you through the architecture in great detail. Thanks a lot. You guys have a great week. That's it for now.
