AI change management: three principles for an AI rollout your team will join

Transcript

Your AI engineers probably hate that you’re doing AI. And I say that not because I have some secret insight into your company. I say that because when I talk to people, right, whether they’re leaders, whether they’re individual contributors, what I hear really consistently in any team over, say, 50 is that you really do have a collection of folks in your business that are not fans of AI at all and not happy that you’re doing AI, actively resistant. In fact, there’s a survey that came out of global companies that showed that a third of employees globally admit to sabotaging AI. So this is a very real issue. I want to talk today about how I address that concern in AI rollout.

And so I’m going to lay out for you what you get in this video. Number one, we’re going to talk with leaders together. Even if you’re not a leader, you want to be in on this conversation. This is what I actually tell leaders. What, what do you say? How do you start to roll out this change?

What is the commitment you need to make? Number 2, we are gonna talk about how you scope and actually set a parameter for starting your change in AI and what that means and the details. And so people ask me all the time, like, where do you start on AI? That is your answer. You’re gonna find out. And why, and why that affects our whole team, why that affects everyone here who’s listening when they go through AI transformation.

And number 3, we are gonna make sure that we talk through some of the really messy details when you go from it’s a cute little pilot to this actually has to scale. What does that mean for people? What does that mean for our systems? How do we have those work together more productively? And I’m going to give you the, like, the secret sauce all the way across, right? I’m going to tell you what I actually am hearing from people, what I’m hearing from leaders, how I’m seeing the pitfalls happen, and what I do to go around them.

So I hope you have fun with this one. Think of it as like a sneak peek behind how I talk with leaders about this complicated issue, and I wanted everyone to hear it, so I’m sharing it here.

And we’re going to start with number 1, which is the contract that you have to effectively sign up for with your team if you want them to come along for the ride with you. And I say contract not because it’s necessarily legally binding, but it is, it is a commitment. It is a commitment that you will need to make publicly as a leader if you are going through the process of AI transformation, which, let me tell you, the world consists of people who are either AI native in their organizations or people who are trying to be AI native. There really isn’t a corner of the workforce anywhere that I have seen that isn’t at least somewhere on that spectrum. You may be early in your journey, but you’re still looking at AI. There are folks way outside tech who I talk to who are like, oh, you know what?

I used Codex to rebuild a tool in my factory recently. Like, it’s just happening. And so in that world, what we really need is a commitment from leadership that they can stick to that reflects their commitment to the wellbeing of the team as a whole. Because if you are asked to participate in an AI rollout. And you look at that AI rollout and you’re like, I feel like this is targeted at me. I feel like this is coming for my job.

I feel like all they want to do is cut headcount. I encourage leaders when they are facing that situation to just very simply address the elephant in the room. I think so much of the time, if we aren’t addressing the elephant in the room, people just aren’t able to move forward. And so I say, look, you can be really honest. A lot of AI coverage, especially in news media, is around AI and jobs. Now, there is no consistent evidence still at a national scale in the United States, which is the most farthest along in its AI adoption journey, that we have any impact on jobs.

It doesn’t matter because if, you know, Jack Dorsey says he’s cutting thousands of people because he wants to focus on AI, well, people hear that message. People hear, yeah, okay, maybe in the national aggregate statistics it doesn’t matter, but for me it matters. For my ability to put food on the table, it matters. And they need you as a leader to be clear and say, what we are doing here is not designed to destroy jobs in this company. It is not designed to take away your role. Now, I have seen leaders do that in a way that reflects integrity.

I’ve seen them do it by saying, we want to keep our company small and lean. And so our plan right now is to reduce the pace at which we grow, or our plan right now is not to hire a lot more. Because we have AI. So there are ways that you can still talk about headcount effects without talking about cutting people who are currently there. When you are an enthusiastic proponent of AI, when you’re able to use AI practically to deliver value, you’re going to go farther in your career here. And if you are conversely resistant, if, you know, God forbid, you’re sabotaging AI, those are going to be things that are going to jeopardize your career here.

And if you don’t have a commitment to protect the people you have, you just don’t have anywhere to start from. And if you want to have a chance to actually engage with the whole business, you have to be able to talk to them about the goals you have for this rollout. Why are you doing this?

And I find it is so much more compelling to use Jensen Huang’s messaging than just about anything else, because Jensen has talked very openly about the idea that he expects tremendous productivity gains from his engineers, but he’s not letting go of any of them. And he’s told leaders publicly who are letting go of their employees that they just don’t have the imagination to use their teams in the age of AI. They don’t have the vision, the boldness to go out there and find new stuff for the company to get into and do. That’s the kind of thing that I find really resonates is if you say, look, it’s my job as a leader to come up with the larger vision for the company and what we’re going after. I see AI as a tremendous productivity enhancer.

And then at that point, you have, you have their attention, right? You have trust because you are then saying, this is not about taking something away. This is about expanding our horizons.

So that’s principle number 1. Principle number 2 that I find that’s really, really critical, because AI is a whole org transformation approach. Like, there, there, at some point, it’s going to come for the tools, it’s going to come for the way you work, it’s going to come for everything. And so if you just start by saying, it’s going to come for everything, we’re going to do AI as a whole, everyone start doing AI. Well, one, people kind of don’t believe it. And 2, people don’t really understand what that means for their day-to-day.

And so you have to start by saying, we are picking a specific angle to get started on AI transformation here, not just the generic, right? We’re picking a specific place. And the way I tell people to pick a specific place is very simple. I say, you need to pick something that actually drives your bottom line. It may expand revenue, it may cut your tooling costs, I don’t know, but you need to find something that has a meaningful impact on your bottom line so you are motivated to pursue it. It needs to be tied to things that AI has already demonstrated to do well.

If you’re doing this, don’t pick a new unsolved problem in AI as your first place to get started. We need to enable them to be more effective. And so for, you know, X routine calls, we’re moving to AI agents. For more complex calls, that’s what our AI agent humans are for. Or they say we want to maintain the human touch. We want humans to talk to humans, but AI is going to be working in the background to go through the CRM, to go through the customer records, to go through everything about our policies so that every call is a fantastic call.

And when you sit down and you figure out that area of impact, and it doesn’t have to be customer service, there are some initial ones that are in engineering because they feel like they have technical skills and all of that to implement. You got to pick that and to talk about it publicly, even if it’s just for that one team. You’re saying to the whole company, we are starting this journey together. This is where we’re starting and why. So maybe we’re starting in customer service. Maybe we’re starting with an engineering project because we want to learn AI-native ways of working.

Those are 2 common ways of, of starting. But either way, you talk about the scope, talk publicly about what success looks like. Success should not just be people use a tool. Success should be something about the way we work changes so that our customer experience gets better, so that we are able to deliver more value to the business, something that’s going to be tangible and meaningful beyond just the activity changing.

That’s really, really important because what we find is that if you just talk about the activity changing, you get into the position Uber got into very publicly earlier this year where Uber said, you know what, we are really regretting the amount of money we are spending on our tokens because they’ve just been encouraging usage like crazy. And that sends a very contradictory message to team members as well, doesn’t it? Because if you’ve been told and you’re somewhat skeptical, use AI, use AI, use AI, and you finally use AI, and then you get told, well, don’t use AI because we have a token budget. Now it’s extra confusing and you just feel like leadership doesn’t know what they’re doing.

You got to pick where you focus, right? You got to pick and scope the nature of the engagement, make sure it’s meaningful. Talk about it publicly to the whole business. Make sure you prepare for it. That’s, that’s sort of budgeting for it, making sure you got the talent on the ground for it.

And the talent, I want to give you a special note on that one. You need to pick a spot in the company where you have a passionate supporter of AI in a management or middle, middle leadership role. And they need to be able to get excited about this. If they’re not excited, if your team-level leaders are not excited about the AI rollout that you’re doing, the director, VP, all the senior folks, is not going to matter. Like, they can talk all they want in meetings, but if team-level leadership isn’t passionate about this, nothing is getting done. Nothing is getting done.

And so I want to remind everyone who hears this, AI is, at the end of the day, a personal transformation journey as much as a public transformation. And so part of the story that we have to tell as leaders in this conversation is that we are going through this together ourselves. So leaders are needing to learn how to use AI and talk publicly about what works and also what doesn’t.

The third thing that I want to talk about is I want to talk about the ongoing nature of these AI tools and evolution and how you know that you are able to get real value and use that real value to drive sustainable change through the business. And so this is, this is an entangled conversation, but stay with me. AI is evolving quickly. We all know that. I talk about it all the time. AI agents are capable of doing more and more complex jobs.

As they do those more complex jobs, we have to evolve our harnesses to work with them. So that means we have to evolve how we call tools in, in our whole company, that how we, how we use data in our whole company. All of these things that are essentially systems of information processing in the business have to evolve to work with AI.

First, you’ve gotta determine your criteria of success. And this is where I will be really blunt with you. If your first AI rollout doesn’t go well, most people either bluntly persist because they say it’s AI, we just got to get it done, or they take the wrong lesson and walk away from it entirely and just take a break from AI, which I’ve also seen happen. And instead, what they should be doing is they should be looking at specific lessons learned. Was there not a real managerial commitment to AI there? Were team members not given the tools to actually understand how AI works, where the expected failure modes are, how to use AI successfully?

And what success looks like in their role. When you did the technical rollout, did the tool actually do a job that actually helped? I’ve seen so many cases where there’s like Copilot rollout and people don’t use it because it doesn’t actually help. You need to get at the harsh ground truth reality of whether AI is adding value to the business. And so first, find the truth out. Second, once you know the truth, you either need to take learnings away and try somewhere else if the pilot wasn’t successful, which data, tool, and system implications there are for your specific tech stack if you expand what you’re doing to other departments, if you expand what you’re doing to other teams ahead of time.

And the reason why is that people either try and bite off the entire org at that point, which is disastrous because it’s so expensive and it’s so complicated, or they misunderstand what they have to actually address from a tech perspective. You really need to be willing to dive into the technical details on to be successful.

And those technical details, as I’ve been sharing, turn into people details because people are then gonna be like, okay, so suddenly all of our memory files are in Markdown. What does this mean for me? Right? Like, do, do, do I write in the wiki? You have to think about what is the long-term capability that I am evolving into the heart of the business, and how do I make sure that I do it in a way that allows both humans and AI agents to work together? And, and that sentence in and of itself is actually a really compelling sentence to share with everyone.

Major studies are coming out showing that one of the biggest impacts of AI, and it is on jobs, is not necessarily AI killing jobs, although that does happen in places, and I’m not denying. It’s that AI is blurring boundaries between jobs, and that leads to a lot of ambiguity. It leads to a lot of confusion. It leads to a lot of people trying to figure out where their careers are. We either— we need them, which is also true because everyone else is adding them, and so you need them to be competitive, and so the business won’t work without Okay. Or because we have a wider vision, which is always more preferred.

People like vision. People buy vision. People wanna see the wider horizon. And in that world, this is how we’re going to get that done. And you need to tell the story in a way that, that supersedes the technical details.

And so when it comes time to move from the pilot into a larger scale, and you’ve talked through the technical details of what you need to change, when you share that message more widely, talk about the impact to the customer and the benefit to the business that you experience in the pilot. Then talk through at a high level why AI matters to the business as a whole and why you are adding AI agents into the mix. Then talk through the long-term vision that you have for AI agents and how information flows without diving too far into technical detail. Especially in a world after the Hugging Face attack where everyone is asking themselves, what are we doing here? What are we doing here? Are we just adding AI agents into the mix and they’re going to cyberattack other, other companies?

Are they gonna experience cyberattacks themselves? You need to be willing to say, these are the safeguards we’re putting in place to protect you, to protect your families, to protect the companies that we work with, to protect ourselves and ensure that we’re not getting hit by, by cyberattacks or malicious actors from AI agents. This is how we are evaluating AI agent output. These are the, the things that AI agents cannot touch and how we’re protecting them, at least at a high level.

And then at that point, you need to say, and this is what is sacred, right? This is what we are committing to maintaining a human edge and advantage for. And that last part is really important because people will, will jump through a lot if they know where you see the vision for humans in their business. And that’s what I want to focus on for the last part of this video.

We know at this point where we’re seeing a lot of value in AI. We’re seeing it in coding tasks for sure. We are seeing it in the ability of AI to accelerate iteration across the business, whether it’s in coding, whether it’s in writing, whether it’s in mathematics. And we’re seeing it in deep, deep analysis. So can you build me a 20, a 20-tab spreadsheet that analyzes the stock position is something that AI is very, very good at. It’s because what we find really consistently, if you want something professional, you have to have a human go through and take out the LLM-isms.

You have to have a human go through and make sure it actually aligns to the larger strategy, et cetera. Talk about the things that the human is doing.

And, and I will tell you, there are people in San Francisco who will say, no, you should not do this. No, you should not do this because agents are just going to keep magically getting better forever. And if they magically get better forever, then there will be nothing for the humans to do. I don’t believe in that world. I don’t see it. I actually see a lot of changing and evolving roles, but a ton of work for humans.

The people I know who are the busiest are the people who are in AI and around AI, not because they’re necessarily training models, but because figuring out how to work with models successfully is the hardest challenge that I think we’ve had to face in 500 years of the corporation. It’s really, really hard. And so we need to be honest about how engineers are moving toward being system designers. They’re moving toward being— writing evals and actually writing how evals deliver code in a loop so that you can actually push the agent against the standard and deliver working code over time. But it takes work, it takes patience, and it takes talking about the human impacts of what’s going on.

So there you go. Those are my 3 big principles that I walk people through as I start to talk about AI transformation. And, and that is how I address the question that I get. What do you do when people hate AI? What do you do when people in the organization aren’t committed to AI? How do you walk through that process with the business?

I hope this has been helpful. If you’ve had a terrible rollout experience, put that in the comments. And if you’ve had a great one, or if you have things that you’d add to this process, put that in the comments. Thanks, Nate. I’ll see you next time.