Key Takeaways

• Start AI coaching pilots with your most skeptical audience. Corning began with its engineering population — healthy skeptics by nature — who pushed back constructively on both the experience and the thinking. The result: when the pilot was pulled ahead of a full rollout, the complaints flooded in. One engineer’s wife thought he was having an affair because he kept talking about Nadia.
• AI coaching isn’t HR tech — it’s a trifecta. Deploying Nadia at Corning produced three compounding benefits simultaneously: leadership development, stronger employee engagement, and a positive first experience with AI that prepared the workforce for what was coming. Organizations rarely get all three from one investment.
• The ROI conversation starts with what you’re already spending. Jordana’s approach: examine the existing budget for personalized coaching, reallocate a portion toward AI coaching, then build assumptions around engagement score improvement, retention correlation, and productivity impact. Clear assumptions, not perfect models.
• AI coaching scales the bespoke. HR business partners at Corning used Nadia as an OD consultant and thought partner — customizing coaching pushes for sales capability gaps, early career development needs, and succession challenges specific to their business unit. Personalization at scale, without proportionally scaling headcount.
• Measure AI against humans, not against perfection. Jordana’s reframe: the right question isn’t whether AI coaching is perfect, it’s whether AI coaching is more consistent than trained humans delivering the same intervention at scale. The answer, she argues, is almost always yes.
• The biggest opportunity in HR AI is still untapped. Recruiting and talent acquisition have absorbed most of the AI investment in HR. The higher-value work — leadership development, career succession, culture, and performance management — has barely been touched, and those are exactly where AI coaching can have the greatest organizational impact.

———

Full Transcript

Welcome and Speaker Introductions

Alexandra McMurray:  Welcome, everyone. We’re so excited to have a good group of people joining Jordana and myself. We know that everyone is stepping away from something else, and that you’re all incredibly busy — so we really appreciate people taking the time to learn with us, and learn from some of the experiences we’ve had with AI coaching out in the field.

I’ll introduce myself, then introduce Jordana, and we’ll get into some content. I’m Alex McMurray. I’m one of Valence’s founders and lead our client relationships here. I spend my days chatting with people about what’s possible with AI. When we brought Nadia to the mainstream about two years ago, I was having conversations around maybe Nadia would be the AI that a company would buy. Clearly, the conversation today is how might something like Nadia fit into a hugely complex ecosystem of different AI applications — how do we integrate not only with your existing software, but what’s our MCP philosophy, et cetera. It’s certainly been a wildly exciting time, and I am really excited to welcome Jordana, who was one of our foundational clients when she was the CHRO at Corning.

Jordana Kammerud:  Hi, everyone. I’m Jordana Kammerud, former CHRO at Corning. Before that, CHRO at Claire’s and Core-Mark — a lot of different industries. Prior to that, I worked at SC Johnson, American Express, and DaimlerChrysler. I love the Valence team, I love Nadia, and I love even more this conversation about talent infrastructure and the way that AI can enable us in HR to accelerate the things we’re passionate about, which is capability and talent.

A CHRO’s Data Journey Across Three Decades

Alexandra McMurray:  Jordana, before we dig in — you were the CHRO at Core-Mark, then Claire’s, then Corning. What were the big challenges you were seeing across those roles, and how was technology evolving? How did that culminate when you were at Corning?

Jordana Kammerud:  I’ve been in HR for twenty-five years, so I can actually go back to the beginning, when we were all paper — green screens, paging each other. The arc of data transformation and acceleration has been absolutely fascinating in my career.

I’ve also always worked in industries where there were large global populations, multiple languages, very distributed workforces. There was always a challenge in how you consistently approach communication, talent development, and culture — finding the sweet spot between consistency and localization, efficiency and effectiveness.

At Core-Mark, a big distributed organization across North America, the buzzword around that time was “big data.” Articles everywhere about how data could transform the way you do business through insights. But what I started to notice — first at SC Johnson and then as I was entering Core-Mark as CHRO — was that people weren’t really putting data around people as the most important priority. They were flowing toward data around customers and sales.

At Core-Mark, they had an exceptional amount of data on what sells, how their sales team works, and how to support that team. What we were able to do was focus on engagement data and really understand how engagement overlapped with performance. At the time it felt very cutting edge. Now I look back at it — it was cutting edge, but with a very limited set of people data. Maybe some one-time assessment data, a little engagement data. And then rich, rich sales execution data.

At Claire’s, you had massive customer data and a very bifurcated population: longtime employees who were the glue that held things together, and a high volume of entry-level retail workers with natural churn. So you had to be very thoughtful about what people data you wanted to capture — I call it “bag and tag” — and what you wanted to apply against business performance to make good choices about what’s driving performance in types of hire, training, development.

The challenge at Claire’s was tight margins, multiple countries, and the pandemic happening simultaneously. The focus there was on getting the right human capital system globally and identifying which pieces of data could give us the insight we needed. The human capital system we used had advanced significantly and had a great data warehouse — we were starting to marry more things from an information standpoint.

Then entering Corning, almost three years ago — a different position entirely. A very large global organization, at the time fifty to sixty thousand employees, now almost seventy thousand, multiple languages. Coming off a disruptive post-pandemic period, there was a need to double down on engagement and culture. That organization, at a hundred and seventy-five years old, had always done a great job on culture — but there was a real desire to reinvest: how do we listen to employees, help them develop, help them be more engaged, and enable leaders to do that?

At Corning there was a lot more data available — analytics had advanced. You could configure much deeper individual questions with location data and other variables. We ran different scenarios and a lot of regression analysis to get a deeper understanding of where we had opportunity against our business objectives and where we wanted to focus our work. That was especially important because, like many organizations, the HR organization had become leaner during that time. Data is very powerful in helping you prioritize when you’re a leaner team.

And now we can talk about AI enablement — and we’re in an even better place.

How Corning Found Nadia: AI Coaching Meets Culture

Alexandra McMurray:  Maybe tell us a little about what was going on at Corning that made you start thinking about AI coaching — and how you heard about Nadia.

Jordana Kammerud:  Starting from the broadest view: every time I’m thinking about a strategy, you’re getting a big broad view of what’s contextually happening. As I was working with the HR and leadership teams on the people strategy, AI was starting to come to the forefront — it was mid-2023, and people were really starting to think about it, even if they didn’t know what to do with it. I knew this was going to slam into us, and we had to start thinking about how to prepare.

Simultaneously, our engagement survey feedback was telling us that people were fatigued and tired post-pandemic. How do we reinvigorate around our core values — people care, development, leadership care?

Those two forces came together somewhat organically. I was talking about these challenges with a colleague in the industry who said, “Have you seen some of the advancement in AI coaching? It might be an interesting way to develop your leaders at scale.” I hadn’t gone deep on it yet. So I connected with Valence, and we started to explore the art of the possible. And we implemented and executed from there.

What’s great about the best kinds of plans coming together is that you start to see additional benefits you didn’t anticipate. We had this trifecta effect: we were coaching and developing people; we were furthering engagement because people felt great about it and were performing better; and without even realizing it, they were having a great experience with AI — which was opening them up to what’s going to come in the future.

Piloting with Engineers: Starting with Healthy Skeptics

Alexandra McMurray:  One of the things you did was start with who you thought would be your most challenging audience. Tell us about that choice and how it played out.

Jordana Kammerud:  Credit goes to the head of talent at the time, Sonya Fabian, and a long-standing HR business partner, Stacy, who had a really good change management outlook on Corning. Their discussion yielded the pilot group: some of our engineering population. They’re healthy skeptics — and because they’re deeply engaged in their own development, they’re a wonderful group to start with. They’ll challenge the thinking, the approach, and the experience in all the best ways.

We were already planning to sign at scale — the pilot was about how to roll out and capture any experiential issues.

The best story: the HR business partner did a report out on how it was going, and said, “It’s going well because, one, we took it away and got a lot of complaints — everyone was asking where it went. And two, one of the engineers said his wife wanted to know if he was having an affair with a woman named Nadia because he kept talking about her.”

Those two anecdotes were probably the most telling in the beginning. We did the analytical work on positivity and net promoter score afterward, but those moments told us what we needed to know.

Executive Decision-Making: AI Coaching vs. General AI Tools

Alexandra McMurray:  There’s always the challenge at the executive decision-making level. Sometimes it’s Copilot, sometimes it’s ChatGPT. Often people have Workday or SuccessFactors and wonder why they can’t just use the AI attached there. How did you navigate that conversation?

Jordana Kammerud:  Tech stacks are always complex depending on what you already have — you want to be thoughtful about not being duplicative or inefficient. First and foremost, we were focused on security and data privacy. That’s the first question and the first gold-star pass.

The second challenge was around duplication. A lot of early questions: why can’t we just build something like this in GPT? The answer came down to: what do you want to invest in from an upkeep standpoint — parameters, safety, controls — and what do you want to be out of the box and ready to go, so you can focus on getting the impact you’re actually trying to achieve? Trusted conversations, great coaching, better leadership, better development.

It became very apparent when we focused on the world-class coaching that was built into this tool and the guardrails around it. We also did what I’d call a jailbreak test — sent people in to ask questions you’d never want an AI coaching tool to engage on. Every time, the answer was exactly what we wanted. That was very telling. When the boundary was there, it held.

AI Coaching vs. Human Coaching: Asking the Right Question

Alexandra McMurray:  As AI coaching evolves, there’s an ongoing debate about what AI should and shouldn’t do — especially in territory that used to belong to HR business partners and internal coaches. How do you think about that?

Jordana Kammerud:  I find this question fascinating because I think a lot of people approach it as AI versus perfection. I like to approach it as AI versus organic human capability.

Can the AI do it more consistently in a way that we would want? If you trained a human, they wouldn’t always get it right. When you train AI, there’s a much higher repeat pattern of the answers you’d want. It’s not perfect, but it’s more consistent than humans at scale. Isn’t that interesting?

It’s about asking the right question, and then really defining for yourself what you mean by that question. Because the second question I ask is: when would you want it to be a human versus AI? And why? So you can pressure-test that why — is it because you actually need real empathy and you think ethically that’s the right thing, or is it because you think a human will perform better given the multitude of variables?

I do think some of the general-use AI tools have a sycophantic quality that can skew sensitive conversations. Nadia is designed to push back — that’s by design. But that will resolve itself.

The two simple questions I always return to: Is it better than a human, versus is it perfection? And what would you want a human to do, and why?

Alexandra McMurray:  There’s an interesting angle there too — why are we comfortable with human failure and not tech failure? I’ve failed so many times and accepted that in others, but we don’t extend that same grace to technology. The bar we hold technology to is unquestionably so much higher than we’d hold our human people in the same roles.

Jordana Kammerud:  That is very intriguing. Why do we do that? I hadn’t thought about it that way, but you’re right.

Alexandra McMurray:  We’ve all gone to the doctor and gotten a diagnosis that wasn’t perfect, and we’re okay with it. But you hear of one instance of AI in medicine not being 100% — and it’s 99.9 — and we’re not comfortable. We’re culturally evolving toward accepting what AI can do, and that’s going to take time.

Customizing AI Coaching for Business Strategy

Alexandra McMurray:  Once Nadia was deployed and coaching was in everyone’s hands, how did you think about tailoring it to serve not just individuals, but also the direction of the business?

Jordana Kammerud:  This is where it gets really interesting. When you get this in the hands of people, you start to see all sorts of organic development of opportunities. And as any leader providing a resource has to do, you start thinking: where am I okay with creativity and organic development, and where do we want to steer how it’s used?

At Corning, we systematically approached our core values and key processes — performance appraisals, development planning — making sure that the overlay of how Nadia interacted, which is customizable and configurable, was grounded in the company’s values. That was the floor.

On top of that, there was more organic, bespoke tailoring. Each HR business partner now had a partner they could not just use for their own development — they could work with Nadia as an OD consultant, as a coach, think about how to support a central initiative, or push specific Nadia interactions for their organization.

For instance, if an organization had a big sales capability agenda, they could tailor some of that work, push certain capability development, use Nadia as both a thought partner and a push partner. If an organization had a succession gap and a lot of early career people, that business partner could use Nadia to tailor a more bespoke early career development experience. The permutations became almost limitless.

The challenge is always making sure that the work is aligned through the strategy and through the organization — clarity about what’s important, so you’re not getting scattershot work.

Building the ROI Case for AI Coaching

Alexandra McMurray:  HR investments are often held to a very high bar for ROI. For those pre-investment and trying to describe the potential of empowering everyone with an AI coach — how would you frame that conversation with a CFO or COO?

Jordana Kammerud:  There are a myriad of use cases, so you always want to focus on the one that creates the most value for the business. In our case, I was able to look at our overall budget for personalized coaching, and say: if we pare that down a bit and redeploy some of it, we could lean forward on Nadia without needing to build a detailed segment-by-segment ROI from scratch.

On top of that, we built an ROI assumption around engagement. Our proposition: by using this to strengthen career development and leadership — and by measuring engagement score increase as a leading indicator — we know that correlates with better retention, which correlates with better productivity. You can put dollars and cents to that based on your individual business to create the financial case.

You have to make assumptions. Any business investment is the same. Be very clear about what those assumptions are and what other variables might be at play. The approach: here are the actual things we’re going to do. Here are the leading indicators we’re going to measure. Here are the output and tail impacts we’re watching — retention, productivity. Here’s what we think it’s going to look like. We’re going to check it and see.

You can also get more precise — do an A/B test with a given population, give Nadia to one group and not another, and see if they perform demonstrably better on sales metrics, for example. There are a lot of ways to approach it if you have skeptics, or if you have limited resources and want to make sure you’re putting your money on the right thing.

Alexandra McMurray:  The challenge with something like Nadia is that we don’t really know the full potential yet. We’re still in early days, still making assumptions. When Ethan Mollick was on stage at the Valence Summit, he talked about how somewhat misguided it is to measure productivity on things we don’t even know the value of. You can ask Claude to create more PowerPoints — you can make more in a day than ever before. Is that valuable? Obviously not. But Nadia is elevating the emotional intelligence of a broad group of managers, maybe 5% this year and 10% next year. What might the impact of that be in two years on the capability of the organization?

Jordana Kammerud:  There’s an interesting lens there around what productivity actually is — lowering your input and getting a higher yield output. More PowerPoints might be more input, but if the output is actual performance — better sales, product development — that’s what matters. And human capital is harder to measure for several reasons: humans don’t like being reduced to numbers, you have to get alignment on assumptions, and there’s a long lead and tail time between action and results. But when you distill it to the simplest equation: reduce activity, increase impact or outcomes. If you can reduce the number of conversations that have to happen between generalists and employees around development planning, and yet increase the quality and feeling of engagement and development, that is productivity.

Change Management in Manufacturing Environments

Alexandra McMurray:  Going to deployment — specifically to a manufacturing environment where there are safety implications of accessing a tool while on shift or on site. How did change management look there?

Jordana Kammerud:  The most effective change management is when someone you trust is using the tool. At Corning, it varied by HR business partner, but a lot of it was HRBPs walking people through those first passes — getting people comfortable, showing value creation. The second lever: integrating Nadia into a process people already had to do anyway, like performance management. Now life is easier for them.

The simple framework: I know how to do it, and I see value in it — I’ll do it. I don’t know how to do it — harder to get people to engage. I don’t see value in it — not happening. But if they know how to do it and they see value in it, that’s the sweet spot. And the degree to which each organizational area focused on those two things showed directly in their adoption rates.

Alexandra McMurray:  It’s really about who’s role modeling and who’s doing the hand-holding, more than the actual environment people are in — which tracks with data from our clients at Analog Devices. If your manager is using Nadia, you’re fully two times more likely to end up using AI coaching.

Jordana Kammerud:  Yes — and to add one piece: in areas where you have a more distributed workforce or people don’t regularly interact with a device, the access and value-creation question is a real one. It comes down to what electronic flow or device that person is already operating with. And when you see that unlock happen — when the access fits the workflow — engagement follows.

Unstructured Coaching Data and Organizational Intelligence

Alexandra McMurray:  We’ve talked about the era of big data. Right now, you have a huge set of unstructured data in something like Nadia — and we’re just beginning to understand how to draw insights from it. How comfortable are you drawing an insight like, say, attrition risk in North American sales because we’re seeing a lot of stress-related conversations — even when that relationship isn’t yet proven?

Jordana Kammerud:  I’m comfortable — because all data analysis is about statistical prediction. The likelihood of something. None of it’s perfect. It’s all directional. The question is how strong that direction is and how much noise you can filter out.

The challenge is having really good partners who understand data science, who understand what data you’re collecting and could collect, and how to structure it based on the insights you actually want to produce. And having enough of a time horizon to get really comfortable that your prediction is sound.

In the absence of that time horizon, you have a lot of data from other sources that give you directional examples. You don’t have to prove that disengagement in your organization always yields departure — because you have that data so many other places. Unless you’re an extreme anomaly, when you see engagement decline, you will see attrition increase. It proves out almost all the time.

We humans are not as different from each other as we like to think. We’re actually quite similar. There’s a lot of data out there that can tell you directionally where things are headed — you don’t have to generate every data point yourself before acting. But my advice is always: get a really good data insights partner, internally or externally, and work with them to get comfortable with what you have, what you’re comparing it to, what you need, and how to structure your tech stack and processes to generate the data that produces the insights you care about.

Rethinking HR Processes for the AI Era

Alexandra McMurray:  As a CHRO in this moment, what’s one workflow that CHROs should really be focusing on improving with AI?

Jordana Kammerud:  This is the conversation I’m dying to have with everyone on this call. If you look at HR, we’ve done a lot with AI in process and recruiting — talent acquisition. But when you go up the value stream — talent development, people development, talent movement, culture — the work that can have real impact on business strategy in greater ways going forward — we have a tremendous opportunity to rethink the way we do things.

For twenty-five years, I’ve been in HR, and we’re still doing the same processes. We’ve moved from paper to electronics with performance management, appraisals, development planning, and career succession based on traditional job moves. But why — when we now have data and tools that enable us to do things much more rapidly, much more organically, much more real-time, much more bespoke for the individual and the organization?

Take development planning. Do you really want very static development plans sitting in a human capital system? Or do you want this organic, tailored nature with some framework and structure that’s happening all the time — prompting the leader, prompting the person, tracking progress, bringing the history of the conversation back into the flow, and interlacing that with data about how the person is actually working?

I go back to my Netflix analogy. I welcome it. I want it for me, if it helps me develop toward my aspirations — makes me feel engaged, makes me feel like I’m valued, like I’m adding and creating value and purpose. There’s a ton of opportunity to rethink all of these processes, and this is an exciting time to do it.

Alexandra McMurray:  So many processes we’re married to as standard, because they were invented in a time when that was the best we could do. The 360, for example — how unhelpful to do that once every eighteen months. How unhelpful as a reviewer to spend half an hour ranking someone you know well on a Likert scale with one comment box. With AI, you could just have a conversation for half an hour about someone — have the AI double-click and probe on those questions — and have everything compiled at the moment when that person is actually looking for feedback.

Jordana Kammerud:  We carry these processes around even though they’re tattered and broken — like a security blanket, our old blankie that we just don’t want to let go of, even if it doesn’t create the value we seek.

And going further on the 360 example: AI can help challenge the biases. It could say, “Here’s a woman on a male team with a lot of feedback of this particular type. We know from the studies that that kind of feedback is often about the observer, not the observed — instead of asking this person to change, maybe what’s needed is a team dialogue about valuing differences and perspectives.” And by the way, if the feedback was that she wasn’t collaborative, we can see that her email collaboration is much higher than others. So let’s have a richer discussion about what’s actually going on here — not to criticize anyone, but to make sure we’re doing what we’re trying to do, which is optimize output by having all these people perform at their absolute best in a way that feels engaging to them.

Closing Thoughts

Jordana Kammerud:  Keep the dialogue rolling. I love this conversation. There’s an MIT article that came out recently that basically said the time is ripe for real innovation in the HR function — make the change now, and AI is the impetus for it. Let’s do it.

Alexandra McMurray:  Thank you all for joining us and for following Valence on this journey. Coming up in the series: our CEO Parker will be introducing new features in Nadia focused on EQ — helping people understand each other better and work better together. The following session will feature Jennifer Carpenter from Analog Devices and longtime Valence adviser Prasad Setty, who helped build Google’s Project Aristotle and Project Oxygen. They’ll be sharing findings from a research study on a year and a half of Nadia use at Analog Devices — finding a correlation between using Nadia and moving up in performance. And in May, we’ll have Raul Valentin and Tim Hourigan, formerly of The Home Depot, talking about AI coaching in the hands of frontline workforces. Register at valence.co.

‍