How AI Agents Are Reshaping the Recruiting Process

Speaker A: Welcome to another Tech Connects podcast. Uh, today we're going to be talking about AI agents and how the recruiting process is changing. I'm very pleased to say that I have the CEO of Phenom here, Mahi, uh, Bay Reddy, and he's going to describe. Well, please, uh, Mahi, introduce yourself and talk a little bit about what Phenom does and then we'll get into our topic today.

Speaker B: Awesome. Hi, Paul. Uh, it's a pleasure to be here. Uh, Phenom has a simple purpose. How can we help a billion people find the right work? That's the only reason why we exist. Uh, Phenom is an applied AI company specifically focused on HR domain. We constantly think about talent acquisition, talent onboarding and talent growth as three different segments and which particular areas we can optimize, automate which particular places we can augment, which particular areas we can, in the workflow, we can make them intelligent and use agents. Uh, those are the combinations. What we really do by using data and analytics and ontologies in the back end, that's what we are good at. But at the end of the day, we are very simple. We want to make sure companies can hire faster and retain the people and grow them within the company, all by using data. Understanding of individuals, understanding a company, understanding a team.

Speaker A: So I think software, uh, has been applied to improve processes for a long time. Um, I'm a technologist of 30 years, so I've seen this evolve. AI is definitely accelerating things and I think it's changing the way HR teams use software. There's a lot of concern about how AI is actually going to be applied and what the implications of that are. So if you're okay, what I'd love to do is break all that down and let's talk about the stages of hiring, kind of what AI has changed, how HR departments are reacting to that, and then what the implications might be to candidates as they go through that process. So my rough take on this is you've got, you got to attract to candidates, job applies, resumes, you're going through an interview stage, you're onboarding, and then there's actually, uh, changes in the way people show up to work. So let's talk about interviewing. Sorry, before interviewing, let's talk about the apply process. What have you seen change over the last few years about how jobs are advertised and then how candidates are applying? And then how is applying changing what HR teams need to think about that part of the process?

Speaker B: Yeah, the changes are not at an hr, uh, level. It's actually at Three different levels. The changes are happening at an industry level, it's at a location level. And the third one is it's the level of the job. Based on that changes are drastic, uh, dramatic and different. And um, if you look at, we look at jobs in five different zones, zone 1, 2, 3, 4 and 5. 1 and 2 are uh, frontline jobs. 4 and 5 are knowledge workers. What we constantly see is the 1 and 2 jobs has a different flow than 4 and 5. And now you apply industry by industry. Retail in 1 and 2 is different from manufacturing. Manufacturing is different from uh, if you take like healthcare frontline or if you take something to do with hospitality. Now you apply the same thing for four and five knowledge workers. Financial segment is different from health care. Healthcare is different from professional services, professional services is different from tech. So what happens is in general we used to think apply is the starting step. But the starting step is how you attract which Paul, you mentioned before, attraction is multiple segments. It depends on the supply demand of the market. And based on that the way you attract differs by the location and the job in which particular level it is operating and which industries operating. Based on that behaviors are constantly changing. That evolution is what we are really seeing. Because of that, a standard workflow of how you recruit or retain or grow an employee is being actually a standard. For the last 30 years it was standardized by HCMS, ATSS per se. But that particular paradigm is basically falling apart. So what we are observing is uh, industry by industry, job profile by job profile, level of the job and the location actually differs uh, differently for each particular job. And the automation changes, augmentation changes, how you use agents changes. The one thing what we have seen is workflows right now are dynamic. But there is no workflow you can 100% automate, but there are workflows you can automate in the front or you can automate in the back or, or there are workflows you can automate in the middle. Depends on different kinds of parameters.

Speaker A: Do you think it's crude to say that if you break down the job types by your level, um, that the higher volume jobs are likely to be less compensated and as a result of that you can kind of group. You could almost draw a range and say the lower the comp, the higher the volume, the higher the automation. That's possible inside of that um, job recruiting without making big errors. But as you get higher in the knowledge working management, more senior levels, you just need a more personal appraisal too.

Speaker B: It's always like that, right? Like so um, if you're buying uh, let's say um, a book, you don't weed out because it's $10 but if you're buying a car you basically put a lot more effort. If you're buying a house you put everything into it. So it's the same thing. The value is determined for what you're really uh, doing. That doesn't mean people are invaluable. But what happens is in E commerce you can learn a lot. But E commerce are all about buying stuff. But human beings are not about buying stuff. It's about connecting and collaborating and really evolving. So that's where the fundamental change happens. In E commerce what happens is you actually look at a thing whether you like or not. But when you're really working with the people, it's not just you like. They also have to like and they also have to like the teams and the companies. That combination is what you need to bring at every point of time. And how you articulate effectively is very important. And because of that workflows will change. Uh, you need a dynamic, evolving, intelligent workflows which uh, is what is a need of the hour and it's what is the need for the next couple of decades together. In my opinion work is fundamentally getting reset. That means all the data in human capital is going to get reset. That means all the workflows will get reset. That also means all the systems will get reset. That is the opportunity, what we have right now to really rethink about not hundred percent, at least 40 to 50% how we are doing business for the last 30 years after Internet evolution to what is a new evolution with the AI is the greatest opportunity, what we all have.

Speaker A: Yeah, it's interesting because uh, that level one and two, the examples you gave volume retail, people that have to show up physically for a job, um, I think that that automation is validated because at the end of the day you have someone that shows up for the job. You can see when you get to the more knowledge worker levels though, AI has created a lot more complexity because you may be in a situation where you never physically meet the person. Just talk a little bit about what are some of the changes there and risks that AI have brought both on and uh, the side of ah, hr. But also where there's risks for hr.

Speaker B: Yeah in parts there are areas where you really see the knowledge worker, there are problems and the various where you don't see anything. So let's say even in the knowledge worker there are jobs where you should be physically present Nursing is one of the best example. Physicians is another best example. Or if you take research, uh, in terms of pharmaceuticals, it's another example. You cannot do it at home. Uh, but if you really be a programmer in tech, or you are a financial analyst, or you are a marketing analyst, or you are a salesperson, you can do it from home. Those are the areas where things are a bit different. But in those areas right now, what's happening across the globe is the fraud is climbing up because of the Gen AI infrastructure.

Speaker A: Yeah.

Speaker B: Um, more in tech space, more in financial space, financial services space. Uh, we are also seeing a lot in insurance, we're also seeing some in pharma. Uh, so based on this, and this fraud increases, it changes by North America is different from Europe, Europe is different from Asia. And within Asia, country by country differs. How do you apply the fraud agents is a very interesting problem to really look at a given company operating. Let's say they have 100,000 employees, 50,000 in US 25,000 are like, let's say in Europe and the other 25,000 are split in Asia between maybe Vietnam, China and India. All the three areas, all the five areas. The fraud is entirely different. So the way you deploy agents are different, the way you deploy workflows are different. The way you really look at people are different. Uh, so that kind of flexibility is what does the requirement. That's what phenom does. What we really do as an intelligent talent experience is all about how can we make it intelligent but make the experience relevant to the person so that in the back end people can get benefit out of it through automation and augmentation and intelligence, um, all by using agents and workflows. So that is what we're really doing. And uh, we are seeing a lot of benefits. I'll give an example. There is a company, uh, like lrc, they really deployed nurse hiring specifically uh, by using voice bots. So they can really look at a lot of company, a lot of candidates who are really coming at who they are, how they're connected and they can really look at who are the people. The end hiring manager has to talk. We have seen the same thing in Thermo Fisher. Uh, if you really look at how many people you're hiring in a manufacturing unit, you can automate 80% of the use case. And the last 20% of the last mile recruiting efforts is where humans are required. And they have changed their recruiters into relationship managers at manufacturing unit level. And then they help the onboarding process much more effective. We have seen that at CircleCare, which is more about, like this, gas stations across the Midwest, you can see people, how they're getting hired and how you can really build, uh, the process in a much more effective format. But we also see the same kind of examples, let's say in Kohl's in Australia. The, uh, way they recruit is entirely different. They actually call 15 people into a store and they ask them to walk through the store for one hour. Based on that, they'll hire one or two people. Just how they walk. But before that, there is a qualification process. All that is automated. Only the process where they walk together as 10 people with a manager is where the qualification segment is. So each particular, uh, company hires differently. And how do you really learn from it, how you build the data set of how they really hire, what their uniqueness is, what a system has to capture. But in the last 30 years, nobody was doing it. And that's where we came into picture. We learned that using ontologies. So we build these contest graphs which can connect these dots, what a company's, uh, real thought process, why they really look at things in a particular format and how they can connect. And if they can really identify, then we can help them to really, um, automate or augment or make it manual based on what the requirement is.

Speaker A: I read an article in the Economist. It was a couple of weeks ago about, uh, one of the biggest things companies can do to recruit is just standardize interview questions. Uh, there's a lot of interview questions that, uh, people that are experienced in interviewing like to ask in an interview. Um, do you think that phenom will help with that standardization? And then how do you coach the recruiters to understand what data really will help provide, where they've been relying sometimes on their intuition to make those kinds of decisions.

Speaker B: Yeah. So the way I think about is what, uh, are the questions you're asking? Are the behavioral questions? Are they, uh, fit questions, uh, for a culture? Are you really asking about the technical questions? Are you asking about the know how questions? Depends on the person, where they're coming from and what their background. Um, things has to evolve. That's what humans does. Yeah. Standardizing simplifies the process. But standardizing complicates the equation. This, these are not anymore like, uh, SAT tests, Like you have a standardized process to really go through. But these are the questions primarily asked based on what is the biggest equation in recruiting? You're trying to find a fit between a company, a candidate and a team, and a manager. That is what you're really finding. The right fit based on which particular level you're hiring. That fit is much more relevant if you're hiring early stage talent. The fit is not that important at the level when you're really hiring an executive. Uh, if you're hiring an executive, a C level person, what are you hiring for? You're hiring for your weaknesses as other leaders. The second thing you're hiring is you're hiring for the most important thread is are you a fit on a mental model level and a leadership style level. But when you're hiring in early stage, mental models are not that important or leadership style is not that important. So where you're hiring, what you're hiring do make a humongous difference. And the same thing is true for retention and the same thing is true for growth of an employee too.

Speaker A: Now I've got a question for you, uh, uh, which you can say I'm, uh, not going to answer. It's fine. Do you have a favorite interview question that you ask?

Speaker B: Yeah, I do. So one of the favorite questions I ask is what is the proudest thing you did in your life? Uh, and then I have a follow up question based on how they answer. What is the proudest thing you did in your work life? Yeah, that gives me a spectrum of thought about who they are, what they're proud of, what are they connected with, whom they are, uh, like, whom they want to become.

Speaker A: You've been a founder of a number of companies. What are you looking for in that answer that helps you get a sense that they're going to be adding value to your, uh, enterprises?

Speaker B: Do you align to our purpose? Do you align to our mental morals? Do you align to our leadership values? Do you, do you align to our rituals? Do you align to our behaviors? That's what an interview is about.

Speaker A: Let's talk a little bit about, um, AI getting much more sophisticated on the candidate side. So candidates are using it, as you said, for generative AI resume, uh, generation and more and more using it in real time for coaching. During interviews, talk a little bit about how do HR teams think about balancing the tools that people are using in those interviews and what they need on the employer side.

Speaker B: So the best spot to learn this is actually in schools like how are students using ChatGPT and how teachers are evaluating. The same thing applies for uh, recruiting and retention too. So using the tools is not a bad thing. But how are they really making it their own and how they're really applying it in an effective format? Everybody has a lot of tools, but very few people can use the tools to create an impact. And how they're creating an impact in that particular flow is what you need to really understand. And how they created that impact, if they worked before and where they worked and how they worked, if they didn't work, if their early talent, how they created an impact in their own life and what they really mean by what impact is.

Speaker A: Right. I read uh, fnom, recently acquired included uh, to think more broadly about workforce intelligence. Talk a little bit about what you're adding as part of that acquisition to the phenom, um, tools.

Speaker B: So one of the weakest link in hr, uh, uh, is the data quality. Why if you look at an enterprise, HR is not even the 10th most

Speaker A: um,

Speaker B: tech intensive spend department. Because of that what happens is HR gets like a lot more, less budget to spend on tech. But if you compare the data you get in hr, uh, it's at least three to five times more data than any other data set in any other department. You include sales, marketing, whatever you want. Because in HR you have a resume, you have a person, you have a conversation, you have multiple people really talking to them, then you have actually bringing them onboarding. Then they stay in the company, leave the company when they're working. There's so much data, all the data to compute you need a lot more resources but you don't have budget. But you need quality data to do prediction.

Speaker A: Right, Right, right.

Speaker B: So that is the biggest challenge in hr. Ah, uh, what we're really observing because of that, what happens is until the data quality is improved in hr, uh the prediction and automation won't be that feasible.

Speaker A: Right.

Speaker B: My personal thought process, whatever has been done in the last 20 years, those are not the companies who will clean it up because they are habituated to really uh, create the same data again and again and they don't know what the data means because most of the companies are only working with structured data which is what you type, uh, or like uh, what you click. But the unstructured data is what you swipe, what um, you record what you spoke or what you reason. That's an unstructured data. That data right now is 10 times bigger than the structured data. Now how do you combine them so that you can really create a new alternative where you understand people in the overall HR ecosystem in much more effectively is the biggest challenge and the biggest opportunity.

Speaker A: Okay, so we've talked about the how do you get applies, we've talked a little bit about interviewing, let's talk about onboarding, um, and starting. There's been Some reports, uh, around companies who have taken candidates all the way through and the person who shows up isn't the person that was interviewed and applied for the role. How do you think about identity verification now in a way more virtual world?

Speaker B: Yeah, so I'll answer the identity verification and onboarding before that. I want to really address the previous question which is applied, uh, AI, which is what is the company we acquired from? So the primary reason for that acquisition is it's about agents for people analytics and also really building workforce insights where you can take actions. So those are the two major threads why we thought of acquiring because in nature the data is humongous, but the quality data don't exist. Because of that quality data don't exist. You cannot take actions or you cannot turn them into people analytics, which are meaningful. So that is a primary reason why we went after that. And that's why for us, data quality is the most important criteria.

Speaker A: And that people data, uh, is part of performance reviews, workforce planning. What's the core of the value?

Speaker B: So again we look at three different areas. It's about talent acquisition side, it's about how people really came in, whom you hire, whom you interviewed, whom you screen, uh, whom you assessed, uh, what are the hoops they went through, how much time they spent and what is the volume you really get. That's one set of data Onboarding side, what's the time to onboard? What's the steps in onboarding? Which step is actually really taking more time? Which step is really giving a bad experience to a candidate to get onboarded? What is a fall off ratio? That is another set of data. Then you have the third set of data, which is an employee growth data set. How are they performing, what's happening within the company, what is a survey data says, then you can really look at, uh, what's happening in terms of their learning, agility, all that. And what does their career path really look like? What is the need of the companies for future career paths then how do you combine that and really give an analytics picture? People analytics is actually split into multiple different areas and each area has its own focus.

Speaker A: Got it. Thank you. Let's go back and talk about identity across the process.

Speaker B: So fraud is very interesting. As I said before, uh, depends on the location it changes, depends on the level it changes. So what happens is, um, a person who actually got interviewed, who got screened, who got scheduled, uh, who got an in person interview to who is getting onboarded. You have to look at identity from multiple angles. Is the identity same in terms of Voice. Is the identity same in terms of looks? Is the identity same in terms of credentials? Is the identity same in terms of personality? All those things depends on the requirement of that particular customer and the kind of process they followed in their workflow. We actually create the identity what is unique for that particular company, for that particular job. And based on that, we constantly really look for, are we really getting anything outside? What they're looking at is the identity matching at what level? If the identity don't match, then create a red flag saying that you have to validate it. So a human has to intervene and really do a validation. We ourselves won't do, uh, validation actually give a signal to make sure maybe this is something which you should validate.

Speaker A: Okay, I see. Um, and then as a, as I think about being an HR manager and using the tools, what do you think I should be focused on in, in this coming year to kind of hone my skills, ready for AI and the advancements that we're likely to see in the industry?

Speaker B: The most important thing is, um, there are two things we should look at. What is your current workflow process, and how are your team structured? And now, in your workflow process, which particular pieces you can automate? Now, how do you restructure your team? Because you can automate and argument certain spots of your workflow, but certain other areas, you have to really deploy more resources to make sure the risk and performance are really coming through. Uh, and that is the top priority for almost every HR leader in the market to really think through. That is what we call as work redesign.

Speaker A: Yes, work redesign. Um, the other thing that I think we might see is fear of using AI. Um, how, as an HR leader, do I make sure that I'm protecting kind of that human element of hiring and not feeding into the concerns that AI brings?

Speaker B: So that is one of the main reason why I'm saying, uh, AI don't have 100% automation capacity. You have to constantly look at, do you have to bring a human in the last mile or the first mile or in the middle, where in the right spot, you have to deploy so that your human experience is right. And to what extent you will automate is also what you have to think through. And if we can really do that, then we can do effectively. But even if you don't automate, you can personalize for that person. Even a human is talking, that personalization till now is only up to what a human understands. But now the intelligence, what we build can help the person on the recruiter side or a Manager side, to be more accommodative and personalized to the extent the talent on the other side, so that they can clearly understand and articulate what's important, what's right, uh, what's their hopes and dreams are great.

Speaker A: Okay, last couple of questions. Give me a, uh, uh, bigger prediction of how you think all this technology is going to change HR over the next couple years.

Speaker B: So my biggest thought process, applied AI, um, applying AI to HR ecosystem has started from last year much more abruptly because of the Genai infrastructure really popping up. This year it will standardize more effectively, but the next coming years it will really give uh, a new definition to it. But the most important element, what am I really thinking about is work will be redesigned at a fundamental level in a knowledge worker space. And that will have a humongous impact on how work is getting done. And what work has to be done like humans has to do. That combination will dramatically, ah, change and AI will help in identifying them, but also helping them in argumenting them. Um, but that will create a lot more jobs than we have ever witnessed. But that will also eliminate jobs which are already existing. But it's all about, uh, learning Agility is what helps people to have jobs and have a better career.

Speaker A: Yeah, we're at a very interesting point. There's this tantalizing vision of humanoids coming in and helping. I see robots being trained. I've got to say I've always been, um, excited by that. But, um, uh, the reality is that it often takes longer to get these technologies out to really transform workforces. I think AI is driving change faster than we've seen, particularly in the last decade. So it's going to be very interesting couple years in HR and HR tech to see how all this, uh, transforms.

Speaker B: Yep. There is no doubt the transformation is happening. Uh, how fast, how slow, it depends on the company and the leader. Um, but it's inevitable. Uh, and um, it's a great opportunity for redesign everything.