How federal agencies can unlock hidden workforce skills

[00:00] John Gilroy:

Hey, John Gilroy here. The federal government’s got a dilemma. Rapidly changing artificial intelligence and a reduced budget. Today, some solutions from Eightfold AI. Hit the music, Manny.

Welcome to the Federal Tech Podcast, a podcast that connects you to federal technology leaders. My name is John Gilroy and I will be your moderator. In the virtual studio, we have Nav Singh, Chief Marketing Officer at Eightfold AI.

I’ll put a link in the show notes to that website. So, Nav, I was at a trade show maybe two, three months ago, looking at all the booths talking to people. I see Eightfold AI. Well, that’s an unusual name and I started talking to them and they have a unique skill set. And I think maybe four or five years ago, this skill set was just okay. But right today, reduced budget, hard to hire people. You got to do more with less. And so, Nav comes in to the rescue. Is that what you guys do?

[01:02] Nav Singh:

Hey, John, great to be here. Eightfold AI, the mission statement, one of the things that we talk about is the right career for everyone. When our CEO started the company 10 years ago, at that time his main goal was you should be able to get a job based on who you are and your capabilities not based on who you know. That’s the fundamental reason for existence for us to make an impact in the world one career at a time where we can match the real skills that candidates bring to the job requirements so that candidates are happy. They have a job that gives them meaning and impact. Employers are happy because they are able to get the skills that they really need to be successful.

[01:45] John Gilroy:

We have a situation and I think I introduced it where we have some people talking about 8 million zero day attacks every day coming up. So the government’s got this big target on it. People are coming after keeping up and companies are trying to hire cyber security professionals. The government’s trying to hire cyber security professionals. So how can you help a group like the Department of War in this challenge?

[02:09] Nav Singh:

The difference between what Eightfold AI does and what other traditional vendors have done is there are traditionally in HR they’ve been systems of record which is what’s your job what’s your profile how much do you earn what’s your performance been performance rating and so on that’s called system of record right all the database with all the information the difference with Eightfold AI is that we focus on skills over the last 10 years we’ve analyzed 1.6 6 million skills that candidates in the world have the and how those 1.6 million skills have been used in career trajectories over time. So we’ve analyzed over 1.6 billion career trajectories. So that gives us a unique vantage point to see if you have skills A, B and C, are you likely to acquire skill D and E more easily than others?

So if you are looking for cyber security professionals as an example, what are the skills that make somebody become a cyber security professional or get into the training and be able to figure out those concepts easier than others. So we can surface the people who are already part of your workforce who could potentially be upskilled easily to become cyber security professionals because hiring is not easy. There are, I spent 17 years in cyber security before joining Eightfold AI. There are millions of unfilled cyber security professionals in the world. Millions of unfilled jobs. So why not train people upskill people that you already have who have the interest who have the skills and who can be easily retrained.

[03:40] John Gilroy:

I have a friend who’s a healthcare executive and what drives him to beat his head against the wall is that he’ll go to an HR organization and say I need someone with a master’s degree, maybe 10 years experience and knows this this and this and they’ll come back and he’ll find a perfect candidate and he’ll say why didn’t you recommend this person well this person only had nine years of experience so I think many HR people can get these, like they put these blinders on horses maybe they put the blinders on HR people and say well Nav has a three PhDs, but he only has nine years experience in this, so he doesn’t qualify. It’s just it it’s self-limiting, isn’t it?

[04:20] Nav Singh:

That’s right. And I’ve been there. I’ve been a Hiring Manager, hiring for people, and there is an intake form or a meeting that we do with the hiring with the recruiter. And in that meeting, we describe exactly what you said. I need somebody with approximately 10 years experience. the recruiter might see that as a limiting criteria, right?

And in addition to that, there can also be bias a recruiter a human bias which can be unconscious. For example, when we see in a resume that somebody was the captain of men’s lacrosse team that means something, right? That means something about the age, maybe something about their pedigree or something about their income level or their household, right? So what we do is we what we infer from that is only one thing which is college athlete and all the skills that a college athlete brings to bear and can those skills be used in an environment by any federal agency. Right? That’s what we look at. So it minimizes the bias. It also looks at skills rather than all of these very specific criteria that a Hiring Manager may have told the recruiter and the recruiter has kind of almost blindly applied to the candidates.

[05:38] John Gilroy:

So, one of my neighbors here, his daughter is at Penn State and she’s studying cyber security and let’s say she graduates in three years. There’s going to be big tech companies knocking on her door and the federal government may be interested hiring her as well or maybe after she gets some experience. So, how can the federal government compete with these big tech companies or can they or do they have any advantages?

[06:00] Nav Singh:

I believe they do. So let’s just talk about reality first which is there is some friction, process friction slow cycle time when you look at USA Hire and so on right the process and candidate hiring experience is typically described as cumbersome and they get limited feedback back and so very capable top AI talent might self-select themselves out the other part is that there is no easy way for federal agencies and government to really see what skills already exist in my environment and how can I reskill and upskill the workforce based on the gaps that I have to be future ready for for the next 5 to 10 years but at the same time there are some advantages some big advantages one of the biggest advantage is mission plus urgency can be a huge differentiator and you can join the government to make an impact in the world right that’s a huge advantage but that, many times I’ve heard that wartime readiness that framing is compelling for many candidates but only if the agencies can translate those mission needs to clearly defined skill sets and then match based on that very quickly. The second advantage I believe is trust and compliance. So government always looks for responsible AI compliance posture like audits government’s transference transparency right that can be part of how the government persuades skilled talent and so I believe that those things can be an advantage but you have to layer on top of that speed transparency and a modern candidate experience in order for for the government to compete.

[07:50] John Gilroy:

Okay, let’s drive down to the Pentagon, Department of War, and we’re just going to march in there and say, how many people work here? Maybe 35,000 people. Okay? And so, if you take any random group of 30 people in Canada, people in Brazil, any random group of that many people, there’s going to be a diverse skill set. And there may be there may be gold in those minds. There may be hidden talents that people don’t know. And I think that’s what if you talk to people who mine for gold or copper they can see and sift through what’s good and what’s bad. And so I guess you have some story with working with the DoD and finding that unusual skill set that maybe people have overlooked.

[08:32] Nav Singh:

That’s right. And one of the strongest especially crisis play arguments can be redeploying internal talent that is always faster than net new hiring.

[08:40] John Gilroy:

That’s it. That’s it.

[08:42] Nav Singh:

And that’s what we help with. We help you discover what we call undiscovered internal talent. And what that does is that instead of doing that shoulder tapping because I know this person has performed with for me in the past instead of that you can actually can actually see the skills that different people those 35 people 35,000 people bring to the organization and what are the adjacent skills you can either get people who already have the skills or you can be reskilled and upskilled pretty easily.

So that’s what I think, I joined ADEL only six months ago but people before me as well as our federal team they’ve been talking to OPM and this is the concept of a broader workforce exchange where you have open roles across agencies they are available they’re matched against the profiles the skills and the career interests right of existing civil servants and that enables talent to move more fluidly across the government.

So one analogy that I always use is like when a pipe bursts right instead of waiting for weeks to screen a plumber to apply to interview right instead you instead search the skills database maybe there is an person in accounting who has who has done plumbing in the background in the past and could fix the pipe right so that’s about finding skills inside the organization not just limiting your search based on credentials and procs scenes.

[10:06] John Gilroy:

So, what you’re saying is you can apply AI to look deeper into a talent base and maybe find things that other people can’t.

[10:17] Nav Singh:

That’s right. That’s right. And really focus on skills and ultimately what it results in is a skills map. Remember, imagine a readiness dashboard in front of the government agency where they can say these are the skills that are required for the next, 5 years. We talk about space, we talk about cyber security, we talk about autonomous execution in the future with agents. What are the skills that are required? Maybe the skills that are required are human judgment, orchestration skills, the ability to do tradeoffs, the ability to take calculated risks and such skills maybe may have been exhibited in certain projects by certain people and certain candidates who are already inside the organization and AI can surface those.

[11:00] John Gilroy:

So now you and I are both outsiders and so if we walked into the Pentagon and said, “Hey, we think this person can do this because of skills analysis,” they’re going to say, “Nav, Nav, Nav, you don’t understand. I mean, we got to have security requirements here. I mean, the stuff we’re dealing with, this isn’t NAV’s donuts in San Francisco. This is serious high security stakes.” And so, how do you respond to that criticism of using skill-based only?

[11:29] Nav Singh:

Absolutely. And AI cannot grant a security clearance but it can reduce wasted cycles and it can improve routing. For example we believe that there should be some data elements that are first class. So security clearance can be our first class data element which is configurable. So the entire pool of candidates you can very visibly see the security clearance and the expiration date. so that you’re only selecting from those candidates and you always are able to see that configurable element. So and and you can prioritize based based on that.

Thirdly, what I would also say is that we always believe there should be a human review and defensibility. So the broader responsible AI posture, right? Human in the loop decision support. So all of that is important. So even after AI has done all the work with the skills-based transparent evaluation plus configurable elements like security clearance and expiration date in addition to that at the end of the day a human should be reviewing all of those decisions. So AI cannot replace adjudication but it can reduce wastage cycles and it can surface the right candidates for you.

[12:43] John Gilroy:

So, how can Eightfold AI walk into an agency and generate enough trust for them to be able to open up their HR records to a company like yours? I mean, this is kind of like some of the company secrets people have. I mean, hey, and we don’t want to let anyone walk in and do it. So, I think this trust is, I think if there’s AI story to be told, it’s all about trust. And maybe with your story, it’s all about trust as well.

[13:09] Nav Singh:

Absolutely. AI can do a lot of things but at the end of the day if people don’t trust it they will not deploy it and they will not get the maximum value out of it and that’s why many of the projects actually fail. So that’s why we’ve been from the very beginning we’ve been investing in white box or what we call explainable AI.

So the agencies can see not only these are the candidates that match your requirements but also why what are the reasons what are the skills that they bring to the table that match the skills in your open role and that’s why the algorithm is saying these are great matches for you. Second, this is completely configurable. So it is possible for in fact that’s one of the best practices that you do what is called calibration.

The federal government agency or the hiring person can say these are the five people who are performing best in this role today. Infer from this from this pool the skills that are required to be successful in my organization. Or you can also say these are the five candidates that I that I love. these are the candidates that are I don’t have here but these are five candidates who are my kind of perfect dream candidates so AI can learn from them as well so calibration and then explanability of match criteria goes a long way and then we’ve explicitly stated that Eightfold AI uses data submitted by the applicants and anything that we get from the organization our customers we don’t use any candidates public data we don’t scrape data so that goes a long way in public sector scrutiny environments.

The last thing I would say is just certifications. So we have FedRAMP Moderate, DoD IL4, we have ISO certifications, we have whole host of certifications and we have an AI ethics council that reviews our decisions. So those people have been our experience with government federal agencies as well. some of the world’s largest companies both in the private sector are and service provider space are our customers and we work with with federal government agencies today as well. So I think all of those are really required.

Last thing actually I should also add one more thing is bias. AI trained on humans who have some inherent bias is going to be biased. Right? So we’ve spent a lot of time in anti-bias or minimization of bias. Our bias audit sample size which we conduct every year is actually we believe the largest in the industry. So our bias audit sample size is 20 million candidates a year which is huge and that’s where we compare ourselves against foundational models and other AI available in the world to say are we minimizing the bias plus we take steps for example instead of showing the complete candidate name you can just show the initials to say let’s let’s let’s reduce the human bias okay

[16:02] John Gilroy:

let’s try a room in downtown DC at one of these fancy hotels we get 50 federal CEOs in a room and little discussion takes place. So if you could stand up in front of these 50 leaders, what one workforce capability should they fix immediately or what’s the key capability? Is it certificates? Is it training? So what would it be anyway?

[16:28] Nav Singh:

It would be really creating a talent marketplace inside their organization which surfaces skills which also looks for the skills that are required in the job jobs that are open and it does that matching. I believe that if you look at the next 5 to 10 years, if you look at cyber, space, autonomous agents, right, it’s going to require skills that may not be prevalent today. But you may have adjacent skills so that reskilling and upskilling is entirely possible. You just need to identify those people who are willing and able to be reskilled very very quickly.

So I think that’s really important. And what I say is that recruitment, recruiting, reskilling, redeployment, all of these are one integrated loop rather than three different boards. So imagine three disconnected boards, right? A recruiting boat, a reskilling boat, and a redeployment board. Instead of these, imagine you build one integrated battleship that can continuously launch, retrieve, rearm its crew while in motion. That’s what will be required because you don’t know what skills will be required in the future. But if you build this infrastructure, then you can be ready for the future.

[17:40] John Gilroy:

Going to go on an usual tal tangent here. If you look at what books are sold, a popular topic now is World War II. And maybe because a lot of those people have died and they’re summarizing it. And when I think of a contested environment, I think of a maybe an island in the Pacific. Okinawa, a contested environment, huh? But I think in 5 to 10 years, the contested environment is going to go up. It’s going to be space and satellites. It’s also going to go in could be inside a data center. And so the talent you’re looking for in the last five or 10 years is going to transition completely in the next five or 10 years where your mindset it’s almost like take off that baseball hat and put on some hat you never dreamed of before. And so the skill set I think that’s going to apply in the next 5 to 10 years. It’s going to have to be so flexible and almost is it self-learning? It’s almost self-learning because so many new things. I mean you talked about certifications. What kind of certification are we going to talk about in three years? It’s going to be any of these frontier models that are popular now could be replaced. It could be the Sing model that replaces it. It’s just it’s it has to be so flexible now. I mean, how do you prepare for that?

[18:49] Nav Singh:

Yeah. When I was talking to somebody and they were an AI expert from one of the firms, right? Somebody was asking them what inning are we in in a baseball game if we were talking about they said we we still in the dugout. We maybe the national anthem is being sung. So you’re not in the first inning. So it is we’re so early in this whole game, right?

We don’t know how it will pan out. So we right now Silicon Valley things are are are happening here. Even large part of America has not embraced AI. Forget about the rest of the world, right? So there is a long game that is to be played here and we don’t know what skills are going to be relevant. But I think what we do know is things that you mentioned space might be might be something that will be much more important in the future. data centers you mentioned and I’ll say one more thing, in the future the we believe the workforce of the future we call it the infinite workforce is going to be a combination of humans and agents where agents do a lot of the repeatable tasks and they do a lot of those tasks autonomously but then humans set the strategy they set the goals they set they evaluate the work of these agents and fine-tune that work they do trade-offs they think of risk takingaking how much risk is we willing to take and so on what is right in this environment and that judgment right those kinds of skills are important and those skills are not going to go away in fact they are even more important in the future and so I think that’s where again it comes back to do you have the people already in your organization who have those skills who have demonstrated those skills in different projects who you can use for either upskilling or reskilling or many of these skills can be applied as is with with a little bit of tweaking right so I Even though we can’t predict the future, we can look at some of these uniquely human capabilities and double down on them.

[20:44] John Gilroy:

I keep think of the people in Annapolis and West Point. I mean, they’re looking for leadership skills, which is an amorphous topic, but I think that’s going to be a skill that’s going be so important. It’s going to be flexibility and leadership and being able to know when to make the right decision with with conflicting pieces of information. And that’s what a battlefield is, is it’s conflicting information and making the best decision that you can. and I don’t know if AI is going to kind of help us or hurt us in that. See what’s going down the road. Yeah.

[21:10] Nav Singh:

Yeah. One kind of analogy I also use is this this skill thing, right? We it’s like moving from sending out a message in a bottle and hoping the right lifeguard finds it to having a constantly updated satellite map showing the real-time location and certified skill set of every rescue swimmer in the country. So those are two very different things. So in one case you’re just passing the message in a bottle and hoping for the best. In the other you have this red n dashboard and you are able to see the gaps and you are able to figure out how I am going to deploy redeploy the skills that I already have in my my environment.

[21:48] John Gilroy:

So Nav what I’ll do is I’ll put a link to Eightfold AI in the show notes and if you’re walking your dog now you want to pursue this further you can go to Eightfold AI and learn more about your company. But I think you’re right in the middle of the fight here. I mean, we talked about the NCAA athlete. He he’s right in the middle of the fight and you’re right in the middle of the fight, too. And I think it’s going to be a challenge for everyone. But I think the idea of looking at your the 35,000 people on your bench to go back to the baseball, you had 35,000 people on your bench. Maybe someone in there speaks Portuguese. Maybe someone in there has analytical capability that that you can apply somewhere. So, it’s a it’s new. It’s creative. People didn’t talk about this 10 years ago, did they?

[22:32] Nav Singh:

Yeah. And John, I’ll tell you one more thing that people didn’t talk about. Like we launched something called an AI interviewer. So in addition to having the skills inside the you mention I hear many times when I talk to federal agencies, they talk about surge capacity. Yeah.

For a project for that surge capacity, you can’t really have human recruiters take weeks to find the right candidates. We are in a position where we can actually deploy AI for it. You can say here’s an AI interviewer looks and feels like human and can do candidate evaluation 24 by7 and every interview is going to be highly consistent based on the evaluation criteria that you have set. So it’s not like the recruiter is tired at the end of the day and wants to close the interview in 5 minutes or 10 minutes. Right?

I’ve talked to candidates who’ve done the AI interviewer. They said that that has been the favorite round because the interviewer was not in a hurry. they didn’t have to rush somewhere and they felt that it was minimization of bias. So that’s a an AI capability, an AI agent that can help the recruiters, that can help the Hiring Managers to get to that surge capacity with consistency and with all the capabilities and skills that you want them for in the job.

[23:46] John Gilroy:

Good. And that’s I’ll put a link to the AI interviewer in the show notes so people can tune into that. And Nav, unfortunately here we are running out of time. You have been listening to the Federal Tech Podcast with John Gilroy. I’d like to thank my guest Nav Singh, Chief Marketing Officer at Eightfold AI.