Join us for this exclusive live demo showcasing our AI-powered Talent Intelligence Platform.
Register for a live demo →
Powered by global talent data sets so you can realize the full potential of your workforce.
Explore talent suite →
How do you get ahead when talent systems are complex and are made even more complicated by uncertainty in today’s labor market?
Download the guide →
Eightfold AI’s Talent Intelligence Platform is now FedRAMP® Moderate Authorized, meeting the strict security standards required for U.S. federal agencies.
Read the announcement →
Agentic AI, autonomous agents that take on tedious tasks, are the future of HR — but not all agents are equal. Explore our guide to what true agentic AI should do in HR.
See what real Agentic AI looks like →
For decades, we believed organizations could potentially move at the speed of technology. Today, however, technology is evolving so quickly that humans and organizations are running to catch up.
Discover how tech leaders can drive transformation→
The HR and talent landscape is experiencing a paradigm shift, driven by the rise of AI, the push for more connected systems, and the growing importance of Talent Intelligence. As AI capabilities advance and talent needs become more complex, organizations are rethinking how they structure their systems, processes, and platforms. At the center of this shift is Talent Intelligence—a critical foundation for driving more connected, data-informed decisions across the entire talent lifecycle.
In this on-demand webinar, we unpack what Talent Intelligence is, why AI is essential to enabling it, and evaluate how a platform-based approach creates greater strategic value. We then explore how Talent Intelligence fits into the broader HR tech stack, supporting alignment across Talent Acquisition and Talent Management. Lastly, our experts look ahead at the impact of emerging innovations like Agentic AI and GenAI, and how these technologies are not just improving efficiency but fundamentally reshaping workflows, redefining roles, and transforming the future of HR technology—with practical strategies to get started.
Watch David Francis, Practice Leader, Research at Talent Tech Labs, and Michael Dunne, Senior Director of Product Marketing at Eightfold AI, in a forward-looking, actionable discussion designed for HR and talent leaders navigating this evolution.
Why should you watch?
Speakers
Michael Dunne and David Francis discussed the evolution and importance of talent intelligence platforms. Michael highlighted Eightfold’s mission to help individuals and employers find the right careers using AI. David emphasized the need for talent intelligence to collect and analyze people data, enhancing talent acquisition, development, and retention. They noted the widespread adoption of talent intelligence in various industries and the role of AI in improving talent strategies. David also discussed the potential of agentic AI in transforming workforce management and the importance of integrating diverse data sources to inform talent decisions effectively.
Speaker 1 00:00
This Global Head of Research at Talent Tech Labs and Michael Dunne, Senior Director of Product Marketing at Eightfold AI, join us today to dive into “Understanding talent intelligence: What to look for and how to choose”.
Michael Dunne 00:29
Thank you. Good day everyone. Welcome to our session here on talent intelligence, talent intelligence platforms and how they’ve been evolving. It is with great pleasure that I had to have David Francis join us from Tech Talent Labs. They’ve done an incredible amount of research on this area now for some years, and we really look forward to having their insights on a lot of the changes are taking place in talent intelligence, how people are reviewing it, a lot of the considerations, such as skills and beyond, as well as I think might be. On top of a lot of people’s minds is the evolution and how to look at that, whether it’s Gen AI, co-pilots or the big, scary word “agentic”. Little more on myself. My name is Michael Dunne, as mentioned in the introduction. I’m a senior product marketer here at Eightfold. I joined Eightfold in what I call 2019 BC, so that’s before COVID. I mentioned that a little bit for part of a reason. One, just a little background on Eightfold. Our mission is to help everyone find the right career in the world, and to do that is twofold. One is with this assistance to each and individual type of job seeker, but also for those who are doing the hiring, the employer, networks, public sector, et cetera. And I say this is because the focus here is applying AI innovation, AI and a talent platform for supporting those needs, those constituencies. So Eightfold itself has been out almost over a decade. We started up in our fast growing company out of Silicon Valley, they have been supporting a range of constituencies, as I alluded to earlier, it’s not just a candidate, not just employer, not just manager, but also executive contractors and the public sector. And has been able to able to support a number of variety of needs, including partners like hydrogen struggles, who do leadership guidance and support work on skills like from EY, as well as work with Josh Bersin with the Global Workforce Initiative (GWI) of theirs, but also with customers such as the Department of Defense with Gig Eagle in their marketplace for our US Army National Guard and Army Reserve, or OneTen which is there to support the hiring of a million people who don’t have college degrees, but they get them into positions where they could have, finally, family sustaining jobs. Now, what makes me excited about this, and I kind of made the joke about 2019 BC, is we had COVID right after I joined. And probably one of the big things with COVID was a lot of corporate America, a lot of corporate Europe and elsewhere, I assume, suddenly realized they didn’t have a full grasp that they liked about their talent, about how work was getting done, and how they have to reevaluate their talent strategies and look at the tools that they needed. And times have changed since it’s been, like five years since, but as we can see, his talent strategies are increasingly important, and we’re seeing that out there, and I believe that’s reflected also with David’s research there, with talent tech labs, probably with the final slide, I just talk a little bit about myself, as well as the organization, along with seeing how COVID probably changed the mindset of the corporate communities, is that we do see a broad based movement towards adoption, talent, intelligence, platforms. So we’ve seen in industries that you are often fast adopters, like financial services, communications, life sciences, but also in manufacturing, retail consumer goods. We are seeing this being implemented worldwide in 150, 455 countries, often with very big conglomerates. And also we are seeing it again, supporting multiple languages. So people are hiring, people are getting promoted. People are being deployed. People are maybe doing non-traditional forms of work, such as project based, billable work in different languages, whether it’s French in Manitoba, English in British Columbia, Spanish in Spain, German, etc. So we do see this being on a broad front in terms of industries, and we are seeing this as a worldwide phenomenon. Now, part of our charter is for supporting town intelligence. And from our perspective, with talent intelligence and the talents platform, it’s the dynamism that matters. It is having AI insights and automation for understanding the talent based on data from both internal for your organization, data captured from the flow of work, also data from anonymized data looking at from the marketplace, from from organizations and the work that they’ve done, and being able to apply that to all stages of the talent life cycle. So really being put into context, what would be of relevance to a candidate that’s just visiting, say, the site of a large financial services firm, and being able to only have highly relevant jobs recommended to them based on the AI. Similarly, it could be in a very large communications firm where they’re trying to develop new teams for new initiatives and can understand what employees, what kind of path, what kind of critical skills they have, what kind of skills gaps they have in place, and how maybe they go inside, promote up, re skill, or go out to the market or contract. And we see that also repeated in terms of the public sector too, in terms of how you can match, maybe persons who are job seekers, unemployed to employer networks that go work with sites for workforce development, for instance. So for us, talent intelligence very much is the leveraging of AI to keep on top of highly dynamic talent conditions, talent situations, talent markets, and then being apply insights and guidance at scale to, I guess this is probably another buzzword, hyper personalized way that is do it on a large scale, but then the guidances are quite personalized to the consumer of that information. So didn’t want to make it too technical or drag this part out, but I did want to at least introduce myself eightfold what our mission is, what we’re seeing with talent intelligence, and a little bit of our own perspective of what we see with talent intelligence, especially us pioneering AI based talent intelligence platforms. So at that point, I think maybe I will step back and let David come forth and introduce a little more about himself.
David Francis 07:09
Great. Thank you so much. Michael, delighted to be here. Thank you for the invitation. My name is David Francis. I’m the Global Head of Research for Talent Tech Labs. I’ll share a little bit about our company, just to let you know, kind of where I’m coming from. And so you know, you know, as we’re talking through the research and the insights that I’ll be sharing, you kind of have a sense of what’s informing that. So we are a research and advisory firm founded about a decade ago, and I’d say we’re unique in a couple of ways. Our focus is exclusively on talent. Technology spans talent acquisition, talent management and the contingent workforce. And the second lens is, our research is through the lens, and has always been through the lens of the practitioner and so kind of our our mission is to help, you know, organizations improve their talent function and talent strategies across the board through the smart use of technology and the lens with which we take is through the practitioners that are, you know, owning, implementing and driving those, those decisions. There we go. So for today’s agenda, we’re gonna spend a little bit of time unpacking talent intelligence and kind of how we’ve organized this. We’re gonna do a little bit of a primer. And so we’ll talk about what talent intelligence, talent intelligence is, the systems that it affects, and get into use cases. We’ll talk about the components of it and kind of implications for how it surfaces within, you know, in real life, in organizations, technology stacks. And then we’ll end with a little bit of discussion on what’s coming around the corner, particularly around evolution in AI and agentic, and how that kind of relates and marries back. And time permitting, we’ll have some time at the end for discussion Q and A. I will note, if you have any questions as we’re going through here, go ahead and put them in the chat. We’ll do our best to kind of get to them as they pop up, and we’ll have some time at the end as well. So with that, let’s jump in before we actually get into the discussion of talent intelligence. I thought to maybe tee this up with a couple of data points from the macro environment to illustrate kind of where we are, you know, economically or macro, economically, and also to kind of highlight why I think talent intelligence is kind of a good solution for the environment that we are in. First data point here is from our friends at PWC. This is from their 2024, global CEO survey. And you know, in addition to being a beautiful data visualization, illustrates an interesting dynamic that’s happening at many organizations. What this shows is, one of the questions they asked was for, you know, companies that had already reported their primary industry. It they asked, What industries are you expanding into? And the net, net here is, you know, the majority of organizations, more than half, are expanding into sectors that are outside, kind of the primary focus. And so, you know, kind of most notably, if you’ve ever heard the comment that every company is trying to be a technology company, you know, that is certainly kind of born out here with technology being the industry kind of most expanded into. But it’s true across the board. You know, auto manufacturers, or traditional auto manufacturers, are getting into the electric industry. You know, many kinds of, quote, unquote, traditional organizations in mining, for example, are, you know, getting into other parts of the supply chain. And I think the takeaway here, for me, is that across the economy, the velocity of change is accelerating. A lot of that is driven by technology and with, you know, all this change that’s happening, the ability to kind of respond to that change, to plan for that change and to make talent decisions around that change is incredibly important. The next data point is from a survey that actually we ran. So we’re a research organization. We love surveys. We partnered with unleash to survey 280 plus organizations on their talent strategies and investments around AI, one of the questions we ask is, what are your top challenges? And if you look at the top four here, in my view, they’re all challenges that can be directly addressed with talent intelligence type capabilities, so retaining critical talent applicant, quality, the economy or business environment, which is kind of indicative of the uncertainty in the environment right now and keeping up with changing skills. Last data point I’ll call out is that there’s actually kind of an economic facet to all this. And so this is an analysis that was done by McKinsey, which essentially quantified the impact of sub optimal talent strategies. What they found is for the median, you know, the median firm in the s and p5 100. You know, sub-optimal talent strategies basically drive half a billion dollars in wasted opportunity, and they kind of break it out by lack of skills, lack of skills, lack of engagement and efficiency and attrition and vacancies. And so again, there’s a real kind of economic cost to, you know, not having the right talent strategy or and talent intelligence is one of the other tools that can actually help inform better talent strategies across all these dimensions. All right, so with that out of the way, let’s actually dive into the meat of the conversation here. I’ll leave with my own definition. It’s actually quite similar to what Michael presented a bit before that. Talent intelligence is a process of collecting and analyzing people data, which includes skills, capabilities, experiences, even things like ambitions, activities and work output to improve talent acquisition, development and retention, as well as inform better talent strategies. And this kind of illustration on the bottom is suggestive of the kind of driver. I would say talent acquisition is both a capability from a technology, but it’s also we’re seeing increasingly a function that organizations are building inside of their talent teams. And so many organizations today actually have a talent Intelligence Team that, you know, may be staffed differently depending on the organization, but the net of it is there are all these different systems that most organizations have that are generating, you know, fairly rich data. And so you know that the highest level things like the HRIS or HCM, which has kind of your core employee data, recruiting tools like the applicant tracking system or CRM, are capturing data on candidates at the point at which they apply, including, you know, their job history and the skills that they bring. Workforce planning tools. You know, talent marketplaces are sensibly trying to capture the skills that individuals have in order to rematch them. You know, learning systems are capturing, you know, metadata on the courses that they do and the individuals that are actually taking those courses and how they’re performing on those courses. And then, not even shown here, but things like performance management tools are getting data on how actually people are doing their jobs, and whether they’re doing their jobs well. And then there’s also systems of activity, where systems where people are actually doing work, which are capturing the actual work outputs that individuals are doing. And so talent intelligence, kind of at the highest level, to me, is basically taking all of this data which lives in all these different systems, aggregating that together and putting a an intelligence layer on top of it, which is going to infer relationships and help answer questions and inform decisions or specific kind of user use cases. And so you’ve got kind of all this data, you’ve got an intelligence engine that sits on top of it, which is driven by artificial intelligence, and ultimately the output is decisions that are being informed, or user journeys or persona journeys that are being made. Now we’re like, like, Where does this kind of exist? Again, this is our talent acquisition technology ecosystem. And so this is our lens into, you know, all of the different solutions that are kind of available for purchase by talent practitioners. This one’s specific to talent acquisition. The way we organize this up at the top, there are four stages which kind of follow the journey of talent. At the bottom, this kind of color coding, are what we call verticals, which basically means solutions that have related capabilities or kind of some common DNA, but they’re not necessarily directly kind of competitive with each other. And then each individual bubble, or sub vertical, as we call it, our solutions that we’ve kind of defined, and you know, the companies inside there tend to have similar features and functionality, and you know, show up together in competitive RF, T’s, for example. And if you notice the light blue, we actually have a category that we’ve defined as talent intelligence, and so that includes matching systems, labor market intelligence, hiring analytics and social search tools, which are basically people search engines, external people search engines. And the common DNA here is that what these kind of sub verticals have in common is that they’ve each been trained, or most of the companies that are providing solutions in here have been trained on a large data set in order to kind of inform talent decisions, in many cases, in kind of a limited context. Now it’s not quite as simple as this, even though this already looks a little bit complicated, because many other organizations which have kind of a user facing platform that they built in the CRM space, for example, or in the internal talent marketplace for example, also provide talent intelligence capabilities, even though they are, kind of primary offering might be a CRM or, again, you know, a internal talent marketplace. And so when it comes to thinking about, you know, kind of how to put this together in an organizational context, we have a few kind of guiding, maybe principles. The First off is, there’s going to be many data generators, kind of, at your, at your, you know, in your organization. And you know, the data generators themselves aren’t necessarily the, you know, the talent intelligence platform that you want to use, kind of for the organization at large. Kind of key considerations here, I would say, is one, the ability for a system to be able to integrate or read data from multiple systems. And so, again, aggregating data across different point solutions is going to be one of the key, key kinds of differentiators, also the ability to use the same system for our intelligence engine for multiple use cases. And so, one of the, you know, I think most of the organizations that we showed on the previous slide would all describe themselves as, you know, talent intelligence applications. And I wouldn’t necessarily disagree, but what I would say is that in many cases, it’s possible to kind of provide one piece of talent intelligence. And so maybe you’re looking at just as an example, you’re looking at quality of fire. And so you might look at, you know, a tool that can do a look back six months into a, you know, an employee’s journey, and then try to kind of reverse engineer, you know, what are the data points that are going to be suggestive of a high quality candidate for different roles? I would say that’s a great use of talent intelligence, but it’s for that one specific use case. And so talent intelligence, in our view, kind of spans the entire talent function. It shouldn’t be siloed in just talent acquisition or even one part of talent acquisition. It shouldn’t be siloed in talent management or one part of talent management. It really should be driving, kind of talent decisions across the board. And so to the extent that a provider or tool is able to use its engine to inform different decisions, kind of across the talent lifecycle, or different use cases or different personas, or ingest kind of new data and meaningfully make decisions from those kind of the more the better, and then the ability to calibrate organizational context. This is very important. We’ll talk about artificial intelligence here in just a moment, but ultimately, in most cases, there’s not going to be kind of a one size fits all. Every organization is different, every even every team and every individual is different. So to Michael’s point about kind of AI at scale, but then hyper personalizing, I agree with that, and part of that is the ability to actually be able to calibrate or fine tune, you know, how the system is actually working to the specific organizational kind of context and decisions or outcomes that organizations are trying to drive on artificial intelligence. So I’m going to spare us kind of a deep technical dive into kind of the differentiators here, the 70 year history of AI. I will spend a little bit time later in how it’s kind of evolved. But because AI is so important to you, kind of talent, intelligence, capabilities, I wanted to spend just a little bit of time talking about why it’s, why it’s important. And be kind of, what should it do, and probably what shouldn’t it do? What is it good at? And, you know, where shouldn’t it be leveraged? So AI, at the kind of most oversimplified, probably grossly oversimplified, level, are from the simplest machine learning algorithm to kind of the most advanced state of the art, cutting edge, large language models that are kind of on the frontier of research. At the core, all of them are prediction engines. They take data in and then they make a prediction about something. And so AI is really good in inferring relationships and discovering relationships, and kind of learning over time. And so it can dynamically update itself. Even if the algorithm itself isn’t changing based on the data that it’s ingesting. It can learn to get better over time. And so one of the things that it does is, you know, over time, systems that are driven by a kind of tend to get better with more use. They’re typically pre trained on a large data set and fine tuned on an ongoing basis. And then it’s a decision orchestrator, which, by that, I mean, like AI, is literally kind of the only way that you can actually achieve talent intelligence at scale. And so in our view, there are organizations that have set up talent intelligence function, hired teams of data scientists, and, you know, are querying, you know, different data sets to try and find answers. And that is one way that you could approach this problem. But ultimately, if you want to do kind of this at scale and efficiently, AI is really the only way that you can do it. So what should it do? It does four things. It can predict, it can synthesize information. It can discover relationships that might not be obvious, or, you know, particularly between data points that are so disparate it would almost be impossible to do that. And it can form, and when I say inform, it is distinct from decide, and so, you know, there’s, you know, we’re in employment broadly as a regulated industry. How you hire people in most jurisdictions is a regulated industry. How you make promotion decisions or compensation decisions is regulated. And while technically, it’s possible, if you know an AI is making a prediction, you could just give it the, you know, the right so to speak, to make that decision. And you know, do something. And some organizations do set some kind of risk tolerance or thresholds above which, you know, they’ll, they’ll, they’ll automate some decision making. I would say, as a general rule, we say that AI should be used to inform and not to decide. And so ultimately, when decisions are being made, that should be informed by tele intelligence platform, but the platform itself shouldn’t be making the final decision. There should be a human that’s, in most or all cases, actually making the final decision. The last thing I’ll say is, because this is kind of so critical to the overall kind of success or lack of success, to an application that’s going to be a talent intelligence platform. I would say that assessing kind of the capabilities in this area is really paramount if you’re, you know, trying to make a determined determination on a, you know, a technology that can support your talent intelligence team or your existing infrastructure. All right, I wanted to talk a little bit about the some of the different use cases, so at a high level, you know, the again, at the the 30,000 foot view, is that, you know, talent intelligence is, you know, lots of data, and we’re going to make, you know, talent decisions. Those talent decisions can be, you know, very varied. And the way we’ve organized this, the way it tends to kind of trickle down, or, you know, maybe triple up, inside of organizations is at the persona level. And so the kind of underlying intelligence engine that powers all of these different personas, and the data set that’s powering, you know, all of these insights is kind of the same, but the thing that different, you know, people in an organization want is different. And so there tends to be kind of the front end experience, rapid experience or application is a little bit different depending on the persona that’s being serviced. So, for example, you know, leaders. You know, the C suite is obviously most interested in kind of the the strategic stuff. So like, how do we compare to our competitors or in the right markets? You know, do we have the right skill sets to win today and in the future? We have the right talent strategy, you know, are we optimally using our talents? And so it tends to be, you know, quote, unquote, products that are a little bit more analytics focused, particularly around kind of the comparison between the internal and the external workforce for talent leaders and CHROs and heads of heads of talent, it is kind of one layer down. And so it’s like, what are our workforce needs? You know? How are we going to fill those gaps? How do we become a you know? How can we be an employer of choice? You know? How do we improve maybe, our retention? How do we identify who are the kind of high potential talent, whether that’s internally or externally? You know, how to identify and bring in the right people to the organization. And how can we kind of reskill and upskill our workforce, given changing business needs, and the kind of quote, unquote, products that are here tend to be, you know, things that are more, a little bit more action oriented. And so it’s not just the tools to help inform the decision, but then it’s to actually execute the strategy. And so the CRM or the ATS workforce planning, intelligent recommendations, and then you have kind of the manager and employee experience, and this is, you know, kind of much more human LED. And so for managers, you know, it’s understanding what are the skills that they have, or what skills they need on their team. How are their people performing, and what’s driving, you know, maybe any gaps in performance, how are they going to be able to, you know, can they find people within the organization to fill open roles or projects? How can they help their employees to grow and achieve their career aspirations? And you know, how can we kind of organize the team to optimally work together better? And so, again, from a quote, unquote product perspective, this tends to be things like succession planning, Team analytics, opportunity marketplace or internal talent marketplaces, you know, learning, assignment and learning, and also hiring. And then for employees, generally, employees are interested in, you know, their own development. And so, you know, how do they find the next opportunity? You know, who should they connect with internally that can help them, you know, grow professionally. And how can they best upskill themselves to either, you know, get that promotion that they’re looking for, or navigate and find maybe a new, you know, opportunity internally in the organization. So things like the learning navigator, kind of career exploration, intelligent job search and mentorship, community. What we’ve seen, kind of, in practice, is that organizations there tends to kind of be two different flavors of application. I’ll say that there isn’t, kind of yet. There’s a race right now for solution providers to kind of have all of these experiences under a single umbrella with kind of a foundational layer of AI that powers all of these different personas. I don’t think anyone has fully cracked the nut bet yet. And the way that these tend to be sold is either at the kind of executive level, where it’s informing a kind of strategic decision. Strategic decisions, or at the employee level, where the goal is to power kind of user experiences, to enable employees to find careers or find new opportunities, or managers to better manage their teams. But in theory, the same intelligence engine can kind of power all of these applications. Now, Les, do you think this is all kind of ivory Castle kind of stuff in the sky? Sky, as I mentioned a little bit earlier, we’re seeing many organizations. I would say that talent intelligence is probably one of the fastest growing functions inside of organizations, particularly larger, larger companies. And this is an example from an organization that kind of illustrates, you know, kind of the, you know, from a functional perspective, kind of the goal, and then the design of their talent intelligence team, and what it’s designed to actually deliver on. So the overall mission is the, you know, high level stuff like analyze and translate business needs, identify kind of critical skills and talent gap, support workforce planning, contribute to their annual operating plan, and then develop kind of function specific talent insights. And then it’s informing kind of across the, you know, the talent organization. Again, it shouldn’t be siloed in any kind of one of these areas. It’s informing talent acquisition. It’s informing HR. It’s informing how they do organizational design, and it’s informing how they do strategy. So maybe from kind of a functional perspective, one suggestion I would have, you’re going to need a technology, obviously, to support the, you know, the function. But also I think, you know, it’s worth kind of thinking through strategically, you know, talent intelligence can be 20 different things to, you know, 20 different organizations. And so kind of explicitly defining kind of a mission and purpose for what it is that you’re, you’re, you’re trying to drive, and what are the specific use cases that are going to, you know, or outcomes that you want to drive in your organization. Is a really good first step to kind of creating the overall strategy, which is then going to help drive the decision making down to what you know technology is going to be the right one to support that, that strategy. I want to talk for a moment here on skills. So this isn’t presentation. Wasn’t designed to be a skills presentation, but skills are kind of the underlying, or one of the biggest underlying, you know, data points that informs many of the decisions that are made with these systems. And so we can’t not talk about skills. And my big point that I want to make here is that like, skills are essentially useless unless they live kind of within a broader context. And so like, if you think about like, what is a skill? You are a skill, this skill is basically just a string of text. And so, you know, here, for example, we’ve got research, but it doesn’t really tell me anything. If I was, say, wanting to use a talent intelligence platform to drive better matching for a role that I had or that I was hiring for, and I put research in there, it doesn’t really tell me anything. It doesn’t tell me, is it academic research? Is it market research? You know, what type of firm was that market research done at what level, you know, was a quantitative, qualitative, you know, was it a startup? Was it a huge organization? And so there’s just, like it the matches we would get if this was kind of the only thing we were using to drive that particular, you know, search result would probably be pretty poor. And so ultimately the skills are kind of most valuable when they’re informed by a broader kind of organizational context. And it’s one of the reasons we say that you need to basically integrate with all these different systems that are going to inform this context. And why there’s often a need to calibrate a little bit to make sure that you know the way that these skills are being used are actually relevant to the context in which a decision is being made, and so understanding, say, the industry, that that skill is being used at the type of organization, the experiences that are running parallel, you know the proficiency of that particular skill, and maybe even the you Know aspirations of the employee. And then, you know, ideally, if it was used in an internal context, you’d have kind of outcomes and performance also tied to that in some capacity. And so one of the reasons why AI is so important here is because in one of the you know, maybe the pressure test that you should be using when you’re, you know, talking to vendors, is, to what extent, you know, can we kind of add additional data or context, organizational data or context, you know, to the to the model, and have it be relevant and help inform or make better the decisions that are being made. You’ve got a couple of questions, so I’m going to pause.
Michael Dunne 31:34
We do have an interesting question from one of our attendees, and the questions, I guess, if I can read it to you, is, are there models that help evaluate the data the company has and needs to effectively deliver town intelligence? Everyone wants us, but what does it actually take to make it happen?
David Francis 31:54
That is a good question. But do you want to take a stab at that, or do you want me to?
Michael Dunne 31:58
Yeah, I mean, I don’t want to I think, as we’ve seen in the conference season, we’ve had testimonials of how people could just get started. And I think some of the approaches is to use older language models might be equivalent of starting with equivalent of a data mart or a subset of a data lake. Often would be working with some of the key systems that they have experience with and integrations with. Part of what that is asked is actually a purpose of the talent intelligence platform, type of needs, which is eventually, with AI you’re gonna be working with data, there’s effort then of one managing that data. So there’s some things that come into play in a central point, centralized point, such as running routines to test the data for mitigating bias. So one example of that is, if you use equal opportunity algorithms for managing the feature set, the what goes in that you don’t then distort the data for that gets used for the analysis. I think also there’s models of progression. So a lot of people will be like, well, think small, think big, but start small. And so you might be by use case by use case basis, that people expand what gets covered in that data. And that data work in the talent intelligence platform. It might be the classic talent leaders had budget, so we went with ta first, and then slowly go, well, we also want to do entire inter, sorry, internal hiring. But then there’s also the needs of, what is some of these benefits? Of, can we get better understanding of our skills and how the work is done, the context, as you just mentioned, to inform our efforts for going out in the market, for recruiting, and then vice versa. What is it we’re actually seeing when we’re recruiting? To help inform how we onboard people and what how we look at jobs internally. So that was a little bit of my take on it. I don’t know if you have you, I mean, my comment would be so I think you’re, you’re, you’re absolutely right.
David Francis 33:50
First of all, many organizations have, like, a pretty bad kind of legacy data infrastructure, and you kind of need good data to make a decision to garbage, in, garbage, out, type of thing. So there’s a couple of ways to approach it. So to answer your question directly, I don’t know if there’s a company that like they have a model where the first thing they’re doing is kind of like evaluating all data, all data types. What organizations tend to do is a couple of, there’s kind of two approaches you can take. One is, you go through an effort where you’re gonna, you know, try to normalize and aggregate data into some type of a data lake, like a snowflake or something like that in the first kind of exercise is, let’s just get our kind of data house in order. And, you know, just tends to be kind of a, you know, kind of a more it driven initiative with, kind of the support of talent teams. And then once that’s done, then we start looking at other applications. The other approach is that you, like Michael said, you start with, you know, smaller use cases. So instead of going, you know, kind of all things across the organization all at once, you start with a a specific, either specific use case, or an area where, you know, you can get something stood up with enough data, and then, you know, kind of as that system learns and new data is ingested, then it gets better over time. So you either kind of start small, or you know, try your best to start normalizing data. But that is quite a common issue across many, many companies. Great question. All right, this here is just another flavor of, kind of a, so you know, kind of following on the theme of skills and context, you know, kind of what a, you know, kind of a slightly simplified kind of data flow diagram, and you know, decisions that might be made around it, and kind of, where are those skills coming from, and how are they used? And so, you know, in this particular case, it would be, you know, related to, you know, maybe an organization’s job architecture would have some compensation data associated with it. You’d have, kind of, your core set of skills that were associated with the role that ideally have been, again, calibrated by some person on the team to make sure that they’re actually fit for that specific role. You might have skills that were parsed from the job description, or maybe that were pulled from an external data source that’s looking at labor market data on an ongoing basis, and maybe even pulling in kind of internal competencies that you might have in your organization as well, when it comes to kind of the specific use cases around which, you know, organizations are using skills. We created this chart here, which, I think it’s kind of a nice visualization. This isn’t, I want to be clear that this is like, super scientific. It’s more supposed to be kind of illustrative, and to give you kind of our read on the market for where, you know, skills based strategies and technologies have been successful at organizations, and where maybe they haven’t been as successful. So the way we kind of organize this on the kind of x axis, you’ve got, kind of point solutions versus more organization wide, integrated solutions. And then the kind of Y axis is, you know, kind of the probability of success of specific skills based initiatives. And where we’ve seen the most success is around things like having a skills architecture, which is basically mapping skills to the job. Architecture skills based hiring. Since the data that comes in tends to be kind of the most current, you know, somebody’s resume, a is easily, kind of possible and B, you know, it’s the most current kind of picture of picture of a candidate at that particular point in time with the role, kind of job design, or skills based. Job design, skills based learning and development, most major systems have kind of meta tagging for skills and then kind of a direct ability to connect directly with other kind of systems of activation and then upskilling programs as well. We’re seeing maybe a little bit more challenge as things around gig marketplaces, though not always skills based compensation, skills based workforce planning, which I think is kind of an aspiration, but hasn’t yet kind of been fully cracked, related to skills based workforce planning. You know, skills based people analytics. So maybe the last slide I’ll end with for this particular topic is kind of the again, kind of the overall, overall, overarching kind of goal here, I think, is to kind of increase workforce agility. And so what talent intelligence should be doing is helping inform, again, the Workforce Strategy, design and then kind of the operational decisions that happen kind of at each underlying stage of that. And so for us, when we think about that, it’s understanding work, disaggregating that into tasks and skills, and then redesigning the work, you know, possibly more optimally. And so that might be building the workforce, hiring, you know, that might be reskilling or upskilling. It might be acquiring the workforce, which is going out and hiring people. It could be augmenting the workforce, which might be a contingent relationship. It could be automating the workforce, which, in our view, is going to become increasingly common through different automations, or through AI or agentic AI. Or it could be what we’re calling a bundling of the workforce, which is essentially, you know, reassuring it to different global delivery centers, and in, you know, coming up with kind of a different structure for how that work is actually delivered and so on. The point of bots, this is probably a good transition point where we’ll get into a quick kind of discussion of agentic solution. And so we’re going to talk about agentic AI. This is a relatively new, you know, this, this concept didn’t even exist really a year ago. You know, everything was generative AI. And now we have kind of this concept of agent agentic AI. So give you a little bit of sense about or sense of what it is and kind of how it’s being adopted. This is a like a gross, probably oversimplification and not meant to offend anybody here, but I think is kind of a nice summary of the differences in the types of AI that are available. Historically, everything was bespoke AI, which basically means that you had a company that went out in a trade machine learning algorithm on some specific data set or combination of data sets to solve some specific problem. And so the kind of the market example I use is Textio, which actually I love text do. For those of you don’t know, it’s a tool that optimizes job descriptions to automatically infer and rewrite them, to basically make them more inclusive and to not knock out underrepresented groups. The large language models like open AI and tropic Gemini and deep seek, et cetera, kind of changed the game, because instead of having one specific model, you now have basically a model that’s trained on the entirety of human knowledge. And instead of it doing one thing, it can kind of do anything. It’s a general expert that can be prompted and queried to provide general information. And what this did is it democratized access to AI. So one of the reasons we’re seeing so many new applications and kind of new companies building applications in previously kind of established industries, is because now everybody has access to the greatest AI that literally has ever existed in human history, at an extremely low cost. The one challenge with, you know, large language models as they were kind of deployed in it in an organizational context, is that they the output that you get is text, and text can be useful. And so if you’re writing a job description, for example, you know, a large language model can do that, but it then can’t take action like inside of the system, like a human, so needs to go and actually do some actually do some work. And so the kind of latest evolution is connecting large language models to actually the systems that organizations use, along with some multimodal capabilities for things like voice, video, et cetera, to actually do work that formerly or tasks that were formerly done by humans. And this is not just a talent position. This is kind of across is kind of across the workforce at large. So they have kind of complex reasoning capabilities the context aware, and they’ve got systems access and guardrails and kind of a defined job, so to speak, in order to actually do things. This here is to kind of little bit highlight, again, difference between the two, you know, kind of the two sets of capabilities. How is agentic different? I’ll show this slide here just for a moment, not going to dive into deep the the main point I want to make across here is that there’s, you know, how this is coming to market. You have kind of foundational models on the lowest end, which have kind of given everybody access to AI cheaply. There’s a growing set of what I call AI infrastructure companies, which basically let organizations build agents, whether those are clients themselves, or whether those are application developers, maybe like eightfold or others that are building talent applications. And then you have the agents, which are basically companies that have built these BAC experiences out of applications in order to do different things. And the way that it’s evolved so far is there are AI workers, AI coaching, which are, you know, essentially kind of more of a talent management aspect, and AI recruiters, which take on various tasks inside of the end to end recruiting cycle. We did a survey recently, again in partnership with unleash, where we asked about kind of the state of adoption at organizations for agentic AI, generative AI, or CO pilots, which is how they were, kind of, you know, how this capability was rolled out, you know, earliest the kind of TLDR here is, you know, about 80% of organizations have co pilot capabilities or will have it within the next 12 months. And so generative AI, or enabling employees with generative AI is kind of table stakes at this point, but I would describe it as kind of small T transformation, where what happened was people gave, you know, employees, a tool and said, You go figure this out, decide a use case. And so there has been some productivity gains from this, but it hasn’t been kind of transformed, transformational to like how the organization runs with the type of work that it does, or how it executes on that work. We do see agents is much more or AI workers is much more transformational and disruptive to how organizations get work done. And when we ask that same question, you know, to what extent are you using currently or investing? What are your plans for investment? About a quarter of organizations are have already invested in some type of AI agent to do work formerly done in full or in part by HR talent acquisition staff. And you know, another nearly 40% or above 40% are planning to invest in such capabilities, you know, in the next 12 months. And so what I would say is, you know, kind of looking where the hockey puck is going, is that these capabilities are kind of coming, whether you you know, whether we like it or we don’t. And so the kind of call to action here is an understanding, kind of, what are the use cases that are going to make sense in your organization? How are we going to handle the change that’s kind of associated with this and how to best kind of leverage these, leverage these. And I would, I would say that you know, kind of behooves you to have a an informed opinion on the right use cases and the right kind of rollout strategy to your organization, because, you know, my sense is just going to happen kind of with, with or without the The the support of talent teams, and with that, I will wrap up this section of the presentation.
Michael Dunne 45:48
David, thank you very much. That was pretty impressive. I actually thought when we went over these slides, I was wondering whether we could get done within an hour. So very great information, great cadence we’ve been actually has invoked a number of questions from our participants. So we do have one question that actually goes back to the talk on AI, and then there’s a second question that’s a little bit related, that does look more focused on recruiting. So I’ll go over the first one, which was asking, Can we fully modify the algorithm so as to tailor the jobs back to a more narrow and pertinent town pool, or is this a classic black box thinking like we cannot understand the weighting and the scoring so as to deliver on a validated ranking of a candidate employer fit? And I’d say for the answer to that, at least, my opinion is there’s actually a middle path between the two. So you might not necessarily be inherently changing the algorithm, but you there’s terms, and I think you use the term self calibration. So there might be a set of filters, set of criteria that can then create, create, establish greater specificity to your needs. Now, on the talent pool, interesting enough, if the AI is working properly, the and you’re done your work, in terms of managing the data, your talent pool actually might be more diverse. It might actually open the aperture to a lot of talent that you missed. Then I think the next piece is all right. Well, you have this talent pool, we made these recommendations, what is the transparency and explainability of that? And I think, as David, you might agree that we have that there is provision for that. So there’s a classic case where the AI could lay out why this person might have these sets of skills, why person might have skills. You want to validate why there might be skills actually, there are adjacencies, or there’s inferences made that could say possibly candidate is stronger and maybe has a stronger match score. So there are means, I think, to answer that question. Yes, means to filter down or calibrate, and then two, yes, explainable. AI, is very much a topic that you’ll probably hear from the vendor community talking about, for those serious about working with and utilizing, harnessing, AI, I know if you have a few other thoughts about that.
David Francis 48:10
I would say the, you know, it, in some sense, it depends on, I think, the actual model itself. So some just kind of, if you’re, if you’re use maybe large language models as an example, so they’re probabilistic in kind of, how they predict the next, you know, the next token, or the next word, basically. And so you know, if you give it the exact same prompt, it will give you two different response, like you can’t get the you can’t kind of predetermine. It’s not deterministic. So if you give it the exact same prompt, you’ll get slightly different outputs. Kind of each time that you do it. There are models which can be more deterministic, but in general, whenever you’re looking at kind of a, you know, like a neural net that’s that’s analyzing relationships between many different data points and billions of data points. Kind of there is a little bit of a black box. But what we’ve seen is that the explainability piece in most kind of major vendors that are offering some type of a solution, a depending on the specific like, you can, in some cases, wait kind of what’s important to the decision, like, what’s more important to you? That’s when I talked about calibration. That’s like, exactly what calibration is. It’s saying, like, what’s important to me, and, you know, what do I want to wait more in a decision or an outcome that I’m that I’m trying to drive? And so that’s definitely possible. And there’s usually just a kind of little user interface where you can do that. And as far as explainability, again, most major vendors do have the capability of looking into, and, you know, essentially showing in, for example, in a match score, kind of, what proportion of the match score was driven by different factors, or what’s the, you know, essentially the decision tree that led to, you know, that particular that match score great.
Michael Dunne 50:04
And then, as I said, there’s a related question more towards the recruiting. And so our participant was asking, Are there any best practices for applying AI common intelligence and recruiting where we deal with a lot of external, unstructured, confidential and personal data? And I’d actually, would probably point out, from my perspective, from my perspective, AI and its ability to kind of go through unstructured data was one of the values of that. So that was one of the advantages with having a solid AI platform. I would also say that, yes, there are best practices in terms of managing the information that goes in equal opportunity algorithms in terms of handling what they call the feature sets, the input, making sure that doesn’t bias the data set. And then also, I think you’d probably want to look to see, are there tools, including ones or machine learning, AI driven tools for handling, addressing PII, personal information. And then also, it was a big deal in the 2019 timeframe to 2023 but there was a question also, how you can anonymize that data for further analysis, and also, especially if your organization was global. So a big thing was about European companies sending information to divisions in North America like the United States, or vice versa. And so you do want to see what kind of work is being done to address anonymization amongst other things.
David Francis 51:30
It’s a little bit of a my thought. It’s a little bit of a technical question, but, you know, this is one of those things you’re going to discover once you actually like, you should like, when you when you, you know, do your RP, or your demo or whatever, you know. Hopefully you’re not finding this out when you’re actually getting the you’re actually getting the implementation. But ultimately, you know, you want to make sure that the system you’re working with can gate the you can gate which data that are actually informing which, which kind of decisions. Is there certain you’d like, if you use perfect example, you know, there’s some things like you’re kind of just either not allowed to do, or some organizations are not comfortable with, with doing with with candidate data. And so you just need to make sure that the system actually has the technical architecture to to make sure that it’s blocked off. And if that’s not feeding the other decision, great.
Michael Dunne 52:16
And then we have another question, still one that is a fan of your evolution of AI Slide, So AI slide, but the question is, are you seeing bespoke in ll model vendors moving towards, I think agentic AI offerings, or will be forced to switch To agentic AI vendors based on particular objectives?
David Francis 52:41
Yeah, so I think the answer is, so there’s, I mean, my sense is that I’m here in the Bay Area, and my sense is there is just like a massive arms race right now to build to kind of win in the agentic space among the builder community. And so what that means, the the the model provider themselves, I think, are, you know, may at some point come out with their their their own agents, and so maybe open AI, at some point is going to start having, you know, ai reekers, AI researchers, or AI software developers. They kind of already have been moving in this direction with, like, their operator product, but they haven’t gone after it as aggressively. They’re more just like that. The infrastructure piece on the kind of builder side, there’s a race between kind of, like, startups that have, you know, basically built a company, you know, started it sometime, like, in the last two years as, like, an agentic first approach to a recruiting solution. And then, you know, like, for example, companies like, hey, companies like eightfold, or, you know, workday, or whoever else that are, you know, already have a talent application and are embedding agentic capabilities inside of their system, whether that is to kind of maximize value of workflows within that system or to connect with other systems. And you know, kind of be a higher level, you know, maybe talent, orchestrator, or whatever. For the, I think, for the near future, like the like, I don’t know who’s going to win in that space. Quite frankly, you know, I think that it’s easier, from a purchasing perspective, to kind of, like, add on to a solution you already have. Then it isn’t necessarily like net new, buy something and then try to integrate it, but, you know, I think it’s too early to tell kind of how it plays out. But certainly there is, there is an arm race. And the options, you know, for from a practitioner’s perspective, you know, the the good news is that it’s, it’s the there are many options, you know, to strategy you could pursue, to pursue, and, by the way, there’s also, you know, kind of the option of building some types of capabilities yourself, which we see many organizations do. In fact, we, you know, just as an example, we did a case study on Walmart, which is one of the early builders and adopters of, like, a, basically an HR agent for HR service delivery that they, you know, essentially custom develop themselves without going to a company that built it and sold it to them. So we have yet to see how, basically, the chips, the chips land at the end of it.
Michael Dunne 55:16
Great, great. Um, I think we have had all the questions submitted right now, it’s come to my attention, but I thought maybe we just step back to about your skills piece, because I think that’s very invaluable, and I think it’s widely seen as skills as a good denominator, common denominator for doing a lot of this analysis. But I think, like with our other discussions, maybe we could talk about things I like beyond skills, like, what about your thoughts on, like, proficiencies or experiences?
David Francis 55:49
Yeah, I think to, you know, again, I going back to the slide I kind of presented earlier. You know, skills are, they’re, you know, without the kind of broader organizational context, I think they’re, you know, more or less useless also, you know, the interesting thing with experiences is, like, the higher you go, depending on the use case that you want, there are some rules that are, like, defined, like, very like, whenever you see a vendor demo, everybody shows you the software engineer, because software engineering tends to have, like, a very specific set skill set. And, you know, they’re coding languages that you have, and there are different libraries you might, you know, have used. But like, the higher you go, kind of leadership wise, like the higher you go up in the organization, kind of the less direct skills matter, and the more you know skills and or you know, competencies and experiences matter. So like, if you run a P and L, have you started a company? Have you sold the company, you know? Have you, have you managed a team? Have you operated internationally? I think there is still a, you know, I think that competencies are basically just kind of a broader, a collection of a broader set of skills. And so, you know, if you’ve managed a P&L, that probably means that you, you know, have you know, team management experience? You have some financial acumen, for example, you know, probably business development skills. And so there is probably, like a layer skills that could go under experiences. The you know, what I would suggest is that the, again, this is kind of an evaluation criteria for, you know, different vendors you might be taking a look at, is, can you take, you know, different data that you might be using already to inform decisions, and add it kind of to a model that’s, you know, skills based or otherwise, and have it understand kind of how those two things relate to each other. And so I think the Heidrick example, you know, you know Hedrick and struggles being, you know, an executive search firm, and you know, has kind of historical data on, on, on, you know, leaders, you know, spanning decades. I think one of the use cases there is they were able to take this kind of, you know, bespoke data set that was custom to them, add it to, you know, the work that you guys did. And, you know, make their talent decisions, you know, even better. And so I think that experiences are huge. And, you know, from kind of a technical perspective, it boils down into, can you take these kind of bespoke data points that might be unique to your organization, which are going to provide more context around the decisions being made, and have they be able to understand the relationship between those and make it better?
Michael Dunne 58:19
Yeah, I agree it’s like skills as a start, skills and contacts, skills are tasks. But if you look at what with the AI platforms, you’re gonna be looking at like, I think you said like leadership. I mean like experiences, especially in leadership search, proficiencies, proficiencies for what you should do when you set your role, role architecture, but then also the proficiencies of folks as you evaluate them for candidates for job. And I also think an interesting thing with AI is being allowed skills and infer skills and then understand potential people. And I think that’s kind of one of the things is a lot of talent gets overlooked. I think Craig lean, who once ran the OFCCP for the United States government under the first Trump administration to point that out himself, that a lot of times can’t overlook. So I think the ability of potential could become important, especially if you’re in industries where maybe a lot of people are retiring, or you’re going through industries where you’re going through rapid transformation, as you mentioned earlier, like auto where suddenly electric electric components for your powertrain become more important. So lot of that going on, I think it’s pretty exciting. I think we have one final question, I think Dave, if we can sneak this in, is, what should HR leaders be asking of their vendors or their partners to ensure they’re getting real strategic value from their tech investment?
David Francis 59:42
Oh, man, what should you be asking their vendors or partners? Or should they get real street value and tech investments? Well, first off, I mean, I think the everything drives ROI for all of these drives back to productivity, and productivity can there’s a there’s a lot of different metrics that impact productivity. I would be asking for real results on, you know, depending the use case, retention metrics, adoption metrics, internal mobility metrics, and kind of improvements to speed and quality of higher and, you know, this is, I’ll just say that the, I’ll leave it there.
Michael Dunne 1:00:36
I think that’s, that’s, that’s first thing is, like, show me the outcome, what I would ask, and certainly the outcomes, I think also your effort, one’s efforts with responsible AI, especially since with regulation out there, it could be increasing, especially if you have operations, say, like in Europe. And then also, I think it is comic record. And then also, I think everyone started on a foot, particular foot, yeah, I’ll say is maybe a red flag to watch out for.
David Francis 1:00:56
If somebody says that it’s really easy, you know, then that’s, that’s kind of a red flag. It’s, it can be incredibly powerful and transformative, but you should go in kind of eyes wide open, instead of, you know, if, if you’re being sold kind of a out of the box, something will be set up in running in a week, and your organization is going to be transformed in a month. You know, that’s probably something to watch out for.
Michael Dunne 1:01:25
Right, like the Ringo star song, it doesn’t come easy. Yeah. And on that note I think we will have our session host wrap up for us, David. Thank you very much for your time and your presentation.
Speaker 1 1:01:39
All right, thank you all for joining us today.