Host Intro (Gene Kim)

Up next is one of the most amazing technology leaders I've met in my career. I learn something every time I hang out with her. And this is, of course, Amy Willard, Global IT Director, Enterprise Core Platforms and Operations, Strategy and Transformation.

I'm so delighted that she's a part of our program committee. And she was the person who told me, "Oh, we all got rebooted at the same time." And so she's going to tell us about her role helping lead AI transformation, not just in development, but across all of John Deere.

This is one of the most iconic industrial companies — 350,000 connected devices, factories on every continent, over 50,000 employees. And so when we were hanging out at Martin Fowler's event, we were just marveling at the speaker lineup today. You're going to be hearing from folks at Netflix, folks at Block, just showing how we are all figuring out this playbook all at the same time.

So I'm so delighted that Amy will be presenting her journey at John Deere. And here she comes. Thank you.

---

Amy Willard

Thank you, Gene. Super excited to be here at the very first in-person AI Summit, and to be able to share a little bit of our story, but also learn from all of you here.

So I'm a part of John Deere, which isn't necessarily seen as a technical company all the time. And so I wanted to maybe start a little bit with showing you our mission and what I wake up every day as a part of at John Deere and trying to accomplish, and why we're so excited about what AI can do to help our customers and help the world.

---

[John Deere brand video plays]

> Do you know where your breakfast comes from? How about your favorite shirt? Or even the road you take home from work? Chances are we had something to do with it. > > It doesn't matter if you've never driven a tractor, dozer, or even mowed a lawn. We work for everyone under the sun. > > We create solutions for folks who work the land and shape the earth. But it doesn't end there. We innovate on behalf of humanity. The machines we make put food on the table. The roads we construct bring communities together. And the tech we create helps us build a better tomorrow. > > We run for farmers, construction workers, and weekend warriors. We run for coders, baristas, and ballplayers. We run for drivers, riders, and bikers. We don't just run for some — we run for all.

---

So that's a little snippet. Thank you, appreciate that.

It's amazing to get to be a part of a company that wakes up and worries about how do we prepare the world for the future — the needs that we, as humans, need to live our best lives on the planet.

And so when I think about implementing AI at John Deere, I want to first maybe give you just a landscape of what we're working with. And I think some of you have similar complexities in your organization.

So we have 70,000-plus employees across the globe. 18,000 technologists — infrastructure, coding, data — somebody who wakes up and worries about doing something with technology each and every day as a part of their role. We have over 70 factories, over 400 offices. We have over 30 warehouses, and we have about 350,000 connected devices across that ecosystem that we need to make sure are active and healthy and engaged in how we want to run our business in the future.

As a part of my organization, I'm responsible for our strategy and transformation org, which I've been responsible for for a while. And in the last year, I took responsibility for what we call our core platform and operations space, which used to be called infrastructure and operations. We rebranded it. The reason we rebranded it — and the reason we put these organizations together — is because I think it is nearly impossible to accomplish a technical strategy or a technical transformation if the platforms that you're using don't make it easy to do the right thing. So the platforms are a really, really important part of accomplishing anything with tech, and putting those in the same organization has allowed us to move faster and think about not only how do our platforms run our operations, but how do they actually create the future that we need within our company.

Which you can see — the top and the bottom of this chart are really the two parts of my role, right? I need to make sure we have operational stability across the ecosystem of John Deere, but also make sure that we're ready for the future. And the future is changing faster than it ever has with AI, right?

---

And if you look at the landscape that we're dealing with — so much change all the time. I really thought it was important to anchor in: what are the three predictions that are driving how we're approaching this at John Deere? And this is an oversimplification for sure.

But we believe — I personally believe and operate with the assumption — that every role in the company will dramatically change and be different in the future.

I believe that the way we run our business will be disrupted.

And specifically for technologists, the function I'm a part of — I believe that the work we do will look dramatically, drastically different than it has in the past.

So just so I know if I'm with people that are similar-minded: how many of you also think this is true for your company? Okay. All right. Some of you. Great. Awesome.

So I could talk about all the use cases we're doing. I could talk about how we're seeing value. I could talk about how we're changing our tech stack — and find me at a break if you want to talk about any of those things.

What I'm actually going to talk about is this.

---

Internally at John Deere, I talk about AI as a bullet train that is coming, and it's going around and around again at a pace that is very, very fast. And we're trying to figure out where's the train going to stop, and how do we get on this train and stay on the train and stay healthy as we do that.

And so I'm really going to talk about our approach to making sure that the organization is ready for the amount of disruption and the amount of change that's coming. I'm going to talk about that from an employee base perspective and from a technologist perspective.

And so one of the things that I wanted to acknowledge with this group — and I've been in the transformation business for a little while — is that this one feels completely different. So how many of you have been involved in one of the transformations on the left-hand side here? All right.

Does the AI transformation feel anything like those? No, right? And so all the tools and tricks that we used for those are helpful, but they're no longer a formula that we can follow. Gene quoted: we all got rebooted. None of us have figured this out. We're just watching everybody around us and trying to solve this as a community together.

And so the innovation is rapid. The experimentation is high. There's no playbook to follow. When we did an agile transformation, lots of other companies in this community — Target and others — we followed their lead, and they were great for us to do so. That doesn't exist today. The technology is changing all the time. The humans involved are anxious, and there really is no defined destination. We don't know where this bullet train is going to stop, and we don't know how long it's going to keep going at the pace that it's going.

And so the question then becomes: okay, what do we do? How do we get ready? Doing nothing is not an option. Standing still is not an option.

And I personally believe that AI transformation is a profound technical change. It's also a profound human change. And so how do we prepare our humans and how do we prepare our technology for the change that's coming — when we aren't really sure what that change is — becomes the question that I wrestle with all the time. And this is what keeps me super excited 80% of the time, and 20% of the time makes me not entirely sure how I'm ever going to keep up with all of this, right? And I assume many of you feel the same way.

---

01Employees

If I think about this from an employee perspective, the real situation that we're dealing with across most companies is: we're trying to bring agents in and trying to bring the humans along, and there's a natural friction in that process.

For us, you can see just a couple of stats to tell you where we are in the journey. Across our employee base, we have about 90% weekly adoption, about 70% near-daily adoption. We've been really focused on that — not because we think knowledge workers using AI for their personal productivity is game-changing value. We're focused on that because we believe that employees using AI each and every day is how they're going to get comfortable with the bigger changes yet to come. Right? And so starting small, getting people comfortable, showing them what's possible — that's really the approach and why this has been so important for us.

I'm going to give you just three takeaways that we've learned in the employee space and in the technology space.

The first one is curiosity. And so we took a strategy that said we're going to provide AI to every employee. There's no way to remove anxiety that we found that was better than providing them the experience to learn and grow and test the technology themselves, right? And so from an employee perspective, making it walk-up easy, making it safe so that they knew what they could and couldn't do and they couldn't really do harm — it was making it safe for them to experiment and have their own aha moments along the way.

And then we also were really clear about what the expectation was: because we're providing this technology, we want every employee to give it a try. We want every employee to lean into the fact that this is going to change our roles, and that the company was equipping you with tooling and training and the resources you needed to be able to start to make this journey with us as a company.

So every employee has access to the technology. Every employee has access to training programs, which I'll talk about in a minute. And so we wanted to make sure that it was really crystal clear that we were investing in our employees so they would invest in themselves.

The charts that you see up there at the top are real screenshots — everything you see here is going to be a slightly anonymized real screenshot from inside of Deere.

The chart on your right-hand side is our adoption numbers. And so we publish this. We publish it down through a certain level in the organization so leaders can understand the adoption in their spaces. They can navigate that. They can make choices. Sometimes we have fun competitions about that — all of those things — but it's big and visible. A similar thing to what we've learned along the way in agile and DevOps with this community.

On the left-hand side is my personal daily AI usage tracker. Every employee has one. We have access to it. Only I can see mine — it's anonymized other than that. You can see where I took spring break, so I'm proud of myself — I actually did not use my agent over spring break. But you can see that every employee, we want them to be able to understand and self-assess: how am I really doing at this? Making it easy for them to do that and have conversations with their managers.

The second thing that I want to talk about is learning. What we learned along the way is that learning programs and adoption programs are not one-and-done like they often are in other ecosystems.

This is a chart of our actual adoption curve. And so we made the choice to not push this out to everybody at one time. We wanted employees to come to the technology, take the training, be engaged, and when they were ready, we met them where they were. What this meant was that we always, over the last 18 months, had people who were on day one and people who were on day 18 months later, right? And so we were constantly rinsing and repeating the early-days training at the same time as adding the 201 and the 301 and the 401 classes — "Agent in a Day" versus "AI 101," right? And the technology has changed.

And so we have a significant number of programs. None of them look like they looked originally. None of them will look like what they look like in the future. But we're constantly adding to that, so our employees always have a place to go learn, a place to go experiment.

And we made a really intentional choice that the AI that we brought to our employees — we want it to be embedded in the work they do, in the applications they use every day. Because we thought that it was much easier for people to understand the change and understand the possibilities in their day-to-day workflow versus: stop what you're doing and go over here to AI, and then AI here, and then come back and do your day job. And so intentionally baking those in, so in each and every experience they're using AI, hopefully just naturally in the work they do.

The last one — and I think you're going to hear a lot of speakers talk about this, John just did a great job talking about this too — is peer-to-peer advocacy. What is clear to me across this transformation is that people trust the people in their own teams far more than they trust a corporate leader trying to drive AI about how AI is going to help them. It's my job to drive this — there's a little bit of bias in the conversations that I have. And so making sure that we have people in the business who are integrated into their organizations helping drive this has been an important part of our strategy.

We have both named people in a formal program, and we have even more people who innovated where they were, and we pulled them into the fold and said, "Hey, you are a natural innovator. By the way, innovation is one of our company's core values. We want to reward that. We want to make sure that you're on a stage and can talk to the people in your organization." And so this was a nomination and a pulling from the community as people stepped up.

And we really did that because this in-context conversation is critical. So how an accountant is going to use AI is different than how an HR professional, and how a technologist is going to use AI in their day-to-day jobs. So we wanted to make sure everybody had a chance to see that from their own local perspective.

---

02Technologists

All right. Technologists — what many of us are.

So we have a bit of a unique challenge as technologists compared to other functions. We are adopting and changing our entire way of working while also being asked to build the things that create value from AI. So how do we do that?

And you can see just a couple of metrics there from our organization. 90-plus percent of our developers use AI most of the time when they code — and I'll talk about why that's a metric that mattered to us. And then we made sure every role in the technology function had a learning program that did continue to evolve. So everybody: here's a tool, here's some programs, we're engaged, and we're invested in making sure that you can make this journey with us.

All right, there's another audience question coming.

So when I first started talking to developers about 18 months ago, the thing I heard almost every time within five minutes was, "Hey, Amy, these are coding agents. You do know that I don't code all day, right? That's not my only job. I code somewhere between 10% and 50% of the time," depending on what metric you want to believe. And so there was a lot of anxiety around oversimplifying developers' roles to just be coders.

And so we decided actively to choose a metric that represented the reality of the role. And this is evolving now — we need to come up with a new metric that is broader than coding moving forward. But what we landed on was not weekly active users, not daily active users. We've never used those for technologists. We've used a metric that is: 80% of the time that you contribute code is AI-assisted. So if you contribute code on a day, did you use a coding assistant to help you? And that's how we got to 90% of the time — if you're contributing code, you are contributing code with AI-assisted. Now, as the whole SDLC is changing, we're going to change that metric. It's more than coding now. But it really allowed us to short-circuit all of the conversation about, "But what about the rest of my job?" or "What about the days that I'm not coding, that I'm designing?" — all of those things.

And we also have a tracker of the days used: when did you contribute code, and did you use AI to do that, at an individual level as well?

As I mentioned before, tooling and training for all the roles is really important to us, and so we continue to provide that and evolve that. It's challenging to keep up with — so some of that's homegrown, some of that is purchased.

And then we also — and I'd love to talk about this on a break with somebody who's interested — made sure we remembered that these developers are a part of a team. So there are experiments related to: how does a team structure work? How do I think about my team differently? How do the roles maybe step into different roles? And how do we learn along the way as we do that as a team?

The next thing — and this one took me a little longer to learn internally — is that it's bimodal in how we're going to have to lead this transformation.

So if you go back in time, many of our internal choices: we would do some proofs of concept, architecture or the platform teams would pick something, and then we would scale that out to everybody. Rinse and repeat. Rinse and repeat. There is no time to do that now in the same way. By the time we finish a proof of concept, the technology we proof-of-concept'd may not even apply anymore. And so we have to think differently.

And so we've taken an intentional strategy to make sure that we're looking at leading innovators and early adopters and continuing to feed them the latest and greatest, because they can handle the churn. They like to be innovative. They don't mind if you change their tech often — all of those things. And many of these folks have been moved into the spotlight. So they've either moved into my team — like Rhodo in the back there, you can wave — Rhodo is a part of my team, he was a part of this activity. Or they've moved into some of our key organizations. They've moved into our key use cases, and so we're recognizing them. We're putting them in a place where others can see them.

While also making sure that we're not disrupting everybody along the way, right? And I'll talk a little bit more about that on the next slide.

We also made sure we set clear expectations: where are we going? What's expected to be a part of our technology function? The image that you see on the right there is a screenshot from our 2025 goal launch session that I led, where we made it very clear — the expectation over the next couple of years, this moved way faster than we thought it would at the time — was 100% of you will use AI as a thought partner. "AI as a thought partner" doesn't really apply anymore. It was our tagline for about a year and a half. "AI as a team member" or something else is probably going to emerge. But we were really clear that that was what we were going to do. We put it on a big screen, and then the next slide after that was, "Here's the programs and how we're going to help you. Here's the tools that we're going to give you." And we also shared that 80% goal as well.

And because I lead platform teams, I'm going to do a quick nod to the platform teams here with this last point. There's a point where we need to make sure that the platforms make the right things easy, right? And so they're ready to scale. That's back to my very early comment: you can't really drive a strategy if your platforms don't make the right things easy. So we have a continuous feed into our platform teams to understand when and how do these things become codified, and they just become easy. And John gave some great examples of how Cisco's doing something like that as well.

All right, and in my last of the two sets of three: experimentation.

As I mentioned earlier, it's difficult to use the patterns we have in the past to decide how do we innovate and how do we scale. It can no longer be done just in architecture teams. It can no longer be done with just the platform teams. It has to be done broadly, and we have to create new ways in which experimentation is okay at the edges. And how do we harness that from the edges and bring that together and decide what to scale?

And so the message we've given internally is: experimentation is now an expectation. Whether you're experimenting in how you do your job — I'm going to code differently today, I'm going to try something agentic, I'm going to do something different — every employee should be experimenting in their role. And in most of our advanced use case areas, they're also experimenting with different tech. So a divide-and-conquer approach so that we can learn as much as we can as an organization and pull that together — which is requiring us to create new communication pathways and rewire how some of those things work, because there wasn't a natural way for that all to feed back into a common learning experience.

But the second point there is, I think, equally true for us. We have to resist the temptation to scale all of the hype. Right? It's super disruptive. So I have 500 teams. If every time there was a new technology that was going to be the thing, I asked all 500 teams to adopt it — that's all they would do. And so making sure that we have this gateway where we say, experimentation is awesome, but are we ready to scale that? And are we ready to put that out to the broad ecosystem and be disruptive to the value we're trying to provide to our customers?

You can see the ratio there of how much I feel like actually makes it through that gate. And this is a very different way of working as a technology function.

And last but not least, the fundamentals still apply. So like most of you, I still have some teams that can't push a small change all the way through in an automated fashion. I have some teams who maybe don't know how to do test coverage correctly. I have teams that maybe don't have their things instrumented with telemetry. All of those things are required for us to be ready for the speed and the pace that's coming. And so we also have ways in which we're trying to scale those fundamentals that maybe we were able to ignore in the past because the constraint was coding, right? When the constraint isn't coding, everything else becomes a constraint, right? And so how do we go back and say what was optional in the past is now required, and how do we scale new ways to close those gaps that we didn't make the time to cover in the past? And so those are the things that are important to us.

---

03Closing

In closing, maybe just three takeaways from my journey and my learning.

The first one is: foundational adoption — the two things I just talked about — is key for this broad organizational impact in the future. So I will say it again: I understand that basic employee adoption is not where all the value sits, but employee adoption is critical. The reduction of employee anxiety, equipping people to see how the world is going to be different, and being enlisted in that — is very important to us in how we accomplish this transformation and how we're ready for that in the future.

Adoption is an evolving journey. So I had to throw away all my playbooks and all my step-by-step processes for how do you train people and how do you drive adoption, because it's changing constantly. And so for me, I have an organization that, generally speaking, goes to market for how do we ready our employees, how do we ready our technologists? And they're constantly evaluating what do we need to do to continue to move this needle and always be fresh and ready to go.

And then, of course, seek value now. So we have high-value use cases. We have bets we're doubling down on as a company. All of those things are very true. But also don't forget that there's so much more to come. And how do we get ready for the tidal wave of what's next versus maybe the tidal wave of what's happening right now? Don't be shortsighted as a leader. Make sure that we're preparing for that.

---

And I've said about five times that I don't have all the answers, so I could use some help. I would love it if you have seen patterns in how you roll out this change, how you ready your organization — how is that working for you? I'm going to steal what we just saw in the last presentation, and I'm sure I'm going to steal more from the rest of the day.

So I would love to learn from you, and if you're curious, I'd love to share more about what we're actually doing in the details. As technologists, if you see things that you think the technology is here to stay on, or we should scale that now, I would love your input there — we're constantly trying to figure out what is ready for us to adopt and share broadly across the organization.

And then I don't think this is the end of the journey. So anybody who's also working on enterprise-level adoptions and implementations, I'd love to stay connected long term to continue to calibrate with each other.

So with that, that's what I have. Thank you.