Recorded at Cornerstone Connect in Chicago, @Cliff Stevenson sat down with @Vivian Jordan of @Cornerstone to talk about how organizations can trust their AI agents. Vivian explains why leaders and frontline employees experience AI so differently, and she walks through what an agent needs to be transparent and trustworthy. It has to see your business context, understand your people’s skills, coordinate with other agents, show up where people already work, and run under strong governance. She also shares Cornerstone research showing that HR and IT have more in common than people think. Both care about skills, time to productivity, and workforce readiness, and that shared ground can turn a turf battle into a real partnership.

Powered by the WRKdefined Podcast Network. 

[00:00:01] [SPEAKER_00] Welcome to the The HR Huddle, presented by Sapient Insights Group, the ultimate resource for all things HR. It's time to get in the huddle.

[00:00:22] [SPEAKER_01] Hello, welcome back to Chicago. We are here at Cornerstone Connect with a very special episode of HR, We Have a Problem. And I am joined by Vivian. No one has name tags on when they come to sit down at the booth. So I'm like, what's the last name? So could you introduce yourself, what you do, and we're going to get into it. I will say before we even get started that as we were chatting beforehand, I was so excited about everything she was talking about that I was like, we got to get this on record.

[00:00:51] [SPEAKER_01] I was like, this is going to be even more off the script than usual.

[00:00:54] [SPEAKER_02] Perfect. Yeah. Hi, everybody. I'm Vivian Jordan. I'm a team lead for product marketing for intelligence, AI and agents. I'm super stoked to be here sharing my observations and thoughts and then some of the insights that I've heard from others and then what we have been working towards, some of the framework that we have tested. So super stoked to be here.

[00:01:16] [SPEAKER_01] Yeah. You work with Anna or at least, yeah. And she was on earlier, not really sure which order I'll go in. But I saw a lot. I'm learning a lot too, right? I think this has been really interesting because we see a lot of the research on, especially around AI, right? Not surprisingly, that's kind of a topic du jour.

[00:01:37] [SPEAKER_01] And there's a lot of education needs to go along with it. But there's also some management of how it makes you feel and what's going on and all these cool things that we could be doing with it. But how do we manage that for ourselves and how we're using it and the workforce? And we're going to get into all of that. So how do you think it's best to kind of jump in? Because I would like to talk a little bit about this way that sort of we feel about AI.

[00:02:06] [SPEAKER_01] One of the things that we've noted in our research is that sometimes the way AI is being used at the leadership level is much different than how it's used for the frontline workers or for the people doing the work. And so there's some misunderstanding of, well, why aren't they using it? What's sort of the pushback? Because they don't realize that it's a different kind of usage models.

[00:02:26] [SPEAKER_02] Yeah. Interesting you said that. That's kind of the lens that we're thinking about AI too, especially the AI and agent in a workforce. And we're thinking that as like two layers in the workforce. You got the intelligence and decision layer, and you got like action layer and admin layer.

[00:02:45] [SPEAKER_02] And really in the intelligence and decision layer, the usage are normally by a more centralized org where say the leaders, CHRO or a transformation leader, especially in the AI age. The lines are blowing a little bit of like AI transformation. It could be from the IT side, it could be from HR side, sometimes it really is a joint effort.

[00:03:08] [SPEAKER_02] And then within that intelligence and decision model, that's where the agents and AI are helping a leader to form a really data-driven, evidence-driven decision that are impacting their people. So normally they're a little bit like top-down and then even including like scenario modeling and workforce planning, right? So things related to that, once you get like important decision made, and sometimes it's about everybody now has like AI mandate.

[00:03:38] [SPEAKER_02] We did a survey and it was a qualitative driven survey for B2B HR leaders and IT leaders. There's 29 out of the 30 that survey from like VP level above are talking about they have some sort of AI mandate. Right. Now coming with that, there's a demand on how are we going to execute? Do we have people to execute? Who are they? How can we move people around? Should we move people around? Should we hire, right?

[00:04:07] [SPEAKER_02] So those are all the questions CHRO being asked by CEO or the board before the next earning call.

[00:04:14] [SPEAKER_03] Right.

[00:04:15] [SPEAKER_02] And so like the business used to operate in quarterly cadence, but they no longer operate in quarterly anymore, right? Like my business doesn't and the product is super fast. And so now you got to get that answer and that like the choice or like decision making between different model, different choices given to you really quick. And that is the intelligence. Yeah.

[00:04:37] [SPEAKER_02] And also the intelligence has to keep up with the speed of business because a skill gap that got delayed for a quarter is a quarter competitor when they have. And like a flight risk identifying like an exit interview is the leader already lost. So like there's a lot of like time they list laid into this intelligence layer. And this is why all of a sudden we talk about strategic work planning and we talk about decision making.

[00:05:06] [SPEAKER_02] In the past, it's always like a long process and you can't really catch up to the business need. Now it's like can't be real time with like agent like synthesizing and capturing all the signals and made it possible. So I really think this is an area where agent is like transforming what the possibility are.

[00:05:26] [SPEAKER_03] Yeah.

[00:05:26] [SPEAKER_02] Now on the second side, we have actions and stuff. So like if we're Cornerstone, right? We love admins and we love our learners and the employees. This is where employee actually feels AI and a workforce readiness in their life because they could be given a really personalized, a personal journey. And that personalization, even though everybody said they do personalized journey, it looked different because it's based on what signal you abstract to.

[00:05:56] [SPEAKER_02] For us, when we have like people graph that abstract to signal from your work, like what do you did in GitHub or Jira or Asana? Like that has project management stuff. You really kind of know what you have already done, what are the gaps are. So the triggered learning or like a performance plan that needed to shift because the goal got shifted. So like those goal agents, coaching agents are all like aligned to there.

[00:06:23] [SPEAKER_02] That's like where employee feels like the action of AI, which is pretty different from leaders. So we're seeing those two major bucket in the workplace where AI is playing a role.

[00:06:34] [SPEAKER_01] That's wonderful. And I think by doing that, by having those sort of actions that are triggered, I think that helps address one of the biggest problems or at least sort of stumbling blocks when it comes to AI, right? Is this lack of trust. That's kind of the word that is unavoidable. And I think by doing that, it helps people understand why, right? Which is really the trust question. Like, why am I being told to learn this? Why am I being told to do this? And they see it happening.

[00:07:01] [SPEAKER_01] I guess it shouldn't be surprising that Cornerstone, of all companies known for learning, would help make sort of frameworks to help understand how that's all it does. Vivian was walking me through some and we'll see about sharing that with all of you. Ways to sort of understand this conceptually, even visually, as we're going through it. Yeah. Maybe you could help if you could.

[00:07:25] [SPEAKER_02] Absolutely. So the way we're thinking about this is really to put it simply, don't let your AI agent be a secret agent.

[00:07:35] [SPEAKER_01] Yes, I love that.

[00:07:36] [SPEAKER_02] We're talking about agents and there were secret agents out there. And then to be honest, an AI agent could very much behave like a secret agent. They go about their routine. They go about what triggers them. They make a decision at the moment. They make a recommendation. You are the one being accountable for the result.

[00:07:56] [SPEAKER_03] That's right.

[00:07:57] [SPEAKER_02] And the moment you cannot explain back to the boss, likely, like, why these people? Why the shortlist? Why is Joe? Why not Anna? You've got to have an answer. And the AI could very much be right. That list could very much be right. But that is not the issue. The issue is you could not explain. The moment you could not explain, the moment the decision got delayed. The moment the decision got delayed, the moment the strategic window closed. Yes.

[00:08:26] [SPEAKER_02] So, like, that trust is so important because it affects all those, like, downstream impact and also how quick you can act on it. And then when we're thinking about it, you essentially have to have four to five components to make an agent transparent, trustworthy, and it's not a secret agent. Especially when we're talking about contacts in workforce place. Number one, the agent needed to see.

[00:08:53] [SPEAKER_02] When we say see, it means that AI needed to understand your context of the business, the context of outside world. And then the industry term for this is called context graph.

[00:09:07] [SPEAKER_03] Right.

[00:09:08] [SPEAKER_02] And then our term for that is our people graph. And then that graph essentially is giving agents a shared understanding of who is who in the zoo, what they do, how they're different, what are the gaps, what are the gaps compared to the outside world, what is the gap between today and tomorrow. So with that shared understanding, you could actually have a multi-agent system that work together.

[00:09:35] [SPEAKER_02] And just like, Cliff, if you and I are in the same org, we have a shared understanding of business goals, like who is who, we can actually work. Yeah. If we don't know anything about each other and we don't have a shared context, we cannot start working right away. Right. So there's no commonality, then there's no collaboration. And with that, so first is a C, but the two is understand. And then in our language, our love language is skills. Of course, yes.

[00:10:03] [SPEAKER_02] And I know Anna talked about the DWTS.

[00:10:06] [SPEAKER_01] Yes, I was about to ask if this is all going to spell something.

[00:10:08] [SPEAKER_02] Yeah, it's all going to, it's all come together. So the DWTS, which is Domain, Work, Tasks, and Skills, ties to this understand where we have the skills engine and the skills architect that really help inference people's skills from the work they do beyond just HRI system. So beyond just like what being recorded last year was being, you know, feedback from your manager and stuff. It's if you did some working asana in JIRA, like that is accounted for.

[00:10:37] [SPEAKER_02] So then you don't have to make this a 12 month data project, which a lot of you guys probably have been through. You're like skills. No, thank you. Bye. And that's what because in the past it's extremely manual. Now you're able to actually abstract all of that with AI a lot faster at skill. Now you actually understand the people and understand what needed to be done. So like that deconstruction of work is critical for agents to have that shared language.

[00:11:05] [SPEAKER_02] Now, the third piece is actually orchestrate. The agents who work with each other want to have shared contacts we're talking about. Two, they needed to have a boundary. So where my work stopped, where your work began, it's where things are getting transferred. And then we also needed to have a structured handoff. So with that contact, with a boundary, with handoff, they can actually coordinate.

[00:11:33] [SPEAKER_02] So for some places, they have a master orchestrator like Salesforce. They have agent course as a massive orchestrator. There's also just like A to A, agent to agent, different places for that single direction work as well. And then the fourth piece is nowadays, a lot of people just don't even have interface with other portals. Like there are people who don't open LMS, Cornerstone. Like I understand. You wanted to just stay in the teams.

[00:12:03] [SPEAKER_02] You just wanted to stay in Slack. You wanted to use cloud. You wanted to use co-pilot, co-work. That's where we need to adapt. Like we got to be where the things are. People actually are doing the work, right? Like a goal, send it back. Like learning, took it from there. So that's the fourth component. You're actually utilizing it. And the fifth component is governance. This is where you got like gravity, privacy, anti-bias, security. Those are really, really important.

[00:12:32] [SPEAKER_02] We are really proud of ours because we have an AI governance body. And then it's very cross-functional. They sit together every week. And they came up with this like beautiful, very thought through governance and how much humans should be in the loop. So like all of those work together. You got agent. You can actually explain what they're doing and why. So I think that is some of the foundation we're considering in terms of don't let AI be the secret agents.

[00:13:01] [SPEAKER_01] I love that. And it's funny, we were talking sort of beforehand in the green room, as I like to say, about, you know, are you noticing any sort of themes? And now that you mentioned data governance has been one. And it's fascinating to think about HR conversations 10 or 15 years ago. And it's like, oh, yeah, you need to know what data governance is. But now it's like, no, you need to live in that world.

[00:13:21] [SPEAKER_01] But I think part of what you mentioned of having these sort of cross-functional, you know, data councils or teams that work across it is it helps address another major issue that's been happening for my eye from the advent of, we'll say, generative AI era. And that is this, what has been often framed as a battle between IT and HR. When we believe, at Sapient Insights Group, and it sounds like you do too, that it really shouldn't be.

[00:13:49] [SPEAKER_01] It should be a collaboration that only works if you're speaking the same language, if we agree on a common set of goals. It sounds like this is sort of addressing that at a macro level while still dealing with the sort of micro issues that come up in your day-to-day work.

[00:14:03] [SPEAKER_02] You absolutely nailed it. I think there were misperceptions probably from both sides. Both sides, yeah. HR people get upset when this become an IT issue to solve because they kind of, according to some verbatim in our own research, they think IT took the people out of the equation. It's all about technology as if it could just solve all the problems.

[00:14:31] [SPEAKER_02] And IT people got upset when HR only thinking about people. And the technology actually is the way to actually spot some problems, to solve some problems. So really, like, we're seeing some misperception. But we actually saw more on, like, the positive future-looking promises. In our survey, and verbatim, there's a lot of, like, oh, we work a lot more with our IT function now. We serve both sides.

[00:15:01] [SPEAKER_02] And then IT be like, we really are brought in earlier. In the past, IT and our survey talk about being brought in only when a system broke.

[00:15:10] [SPEAKER_03] Uh-huh.

[00:15:11] [SPEAKER_02] They're like, we were brought in to clean up the mess.

[00:15:14] [SPEAKER_03] Yes.

[00:15:14] [SPEAKER_02] And then, but now they're talking about they were brought in earlier to, like, putting in the lens of technology and then to work together. And then we're seeing that from HR side as well, where they're trying to make sure the people elements stay constant when we're evaluating a technology. Yes. Now, what we thought is really helpful is if you were going to have those two functions work together, you got to have a shared language, have a shared priority.

[00:15:43] [SPEAKER_02] Like, you got to care about similar things. Right. So you, like, actually worked out. In our survey of more than 100 people of HR and IT leaders, we found they have three things in common that they share. One is a skill optimization for different reasons. HR cared for different reasons than IT, but they both care about it. IT care more about, like, data computability, like how to actually compute the data and so that skills become something they can use.

[00:16:12] [SPEAKER_02] And then the second piece is time to productivity. So obviously, IT needs to move fast. HR also needs to move fast. And then the third piece is this workforce readiness. They all wanted to have the workforce ready to deliver what's next.

[00:16:27] [SPEAKER_02] And with those shared understanding, we have found if we could equip with our HR partner the right framework and language to talk to IT, when the moment they speak the same language and have the shared priority, that collaboration become a lot more natural. And it become the real problem solver. I love it. Because none of each were going to be a blocker towards the end. And then they're all working towards a similar goal.

[00:16:57] [SPEAKER_01] This is wonderful. This is really, I think this is good relationship advice as well. I think the last one ended up being turned up, it seemed like we were going to be talking about football, but it ended up really being about business. And here we're talking about business, but I really think we're talking about marriage counseling. We got to speak the same language. Charitable. Yes, exactly. Exactly. We all want to be in the same place. So all of this is available right now, right? You're sharing these tools. These are all this framework and all of this is out there. See, I love this.

[00:17:25] [SPEAKER_01] I love the idea of taking something conceptual and making it concrete. I've always been impressed by Cornerstone's ability to do that. Thank you. Which I guess makes sense. But yeah, great work on that. So this is really cool. We're talking about things that are happening right now today. Now, what do you think if we're having this conversation a year from now? How do you imagine that conversation might be a little different?

[00:17:49] [SPEAKER_02] Oh, man. Nobody can really see the future. I know. Three months ahead.

[00:17:53] [SPEAKER_01] We used to say three years ahead. We're like, that's ridiculous. No one can do that. Maybe three months ahead. I will take that.

[00:17:58] [SPEAKER_02] Yeah, you can see the field. Like every three months, a new thesis about like what I could do. It's changing, right? Like now the whole like pacing. Let's making sure we slow down. I definitely see this. What we just talked about become even more important, not less. That's why we're like at the beginning, people are just all excited about this car can drive itself.

[00:18:20] [SPEAKER_02] But then you're going to ask, okay, if they're going to go into the ocean and you cannot have the control, who is responsible for that? Right. Like so all of those questions are just going to become even more important. And then the governance piece. We talk about the observability and transparency from like skill being explainable, decision being explainable.

[00:18:43] [SPEAKER_02] But also like behind the scenes, deep inside on the governance piece where whether agent going to have out of box behavior, that become like a little bigger issue. So I'm hoping the industry can really make sure they slow down in terms of having the right guardrail, technology speaking and culturally speaking in place to match the speed of technology development.

[00:19:10] [SPEAKER_02] And I think this is extremely important because HR decisions are high stake decisions. Right. People's livelihood, their careers, their dreams, and all of those needed to be protected and accelerated by technology versus destroyed. So my hope is, you know, those big AI labs to really help with that guardrail. So the model that we're using can actually keep the promise and protect the people.

[00:19:38] [SPEAKER_01] Yeah, I think it was something you said early on, which was that people need to think that they're not just putting in a technology layer, they're putting in a decision making layer. And I think that's good advice.