Designing for AI Impact in Talent

Key Takeaways

• AI adoption fails when organizations lack ownership, enablement, governance, and clear business outcomes
• AI transformation is fundamentally a people and change-management challenge, not just a technology challenge
• Organizations should define the desired business impact first, then identify the right AI solution
• Successful AI adoption requires ongoing experimentation, training, and accountability structures
• Humans should remain in the loop for hiring decisions, even in highly AI-enabled workflows
• AI can improve recruiting quality, efficiency, and experience simultaneously when implemented intentionally
• Operational redesign matters more than simply layering AI onto existing workflows
• Structured AI experimentation creates stronger long-term organizational learning and adoption

Transcript

Note: transcript is auto-generated, there may be typos.

Anita Chandrasekhar: People who can pronounce my entire name right. So I think we’re off to a great start. Okay. So I’m Anita Chandrasekhar, as he said. Um, I lead talent strategy and operations at Zapier, which is essentially the strategy and operations backbone of our recruiting work. And I’ve been at Zapier for a little over four years, and in my role, the cool thing is I get to think about how AI shows up in how we hire and how our team works, which is real fun.

And you’ve already spent so much time today thinking about and hearing about hiring in the age of AI, and meaty topics like candidate fraud, quality of hire. So in this session, what Casey and I are going to do is talk through and bring to life some of those things to something really practical. How do you make sure AI is actually creating impact in your company, and it’s not just a side activity?

And then also, how do you move it from being an org-wide initiative into something that can be real operational leverage within recruiting teams? All right. So with that, here’s where we’re headed. I’ll kick us off with why AI adoption stalls, what Zapier built to fix that, and how it translated into real priorities for our talent team.

And the key thing that I want you to keep in mind in all of this is the thread that ties it together is a simple idea, and you’ll probably hear me say this a few times. You define the impact you need first, and the right tool will reveal itself later. Just keep that in mind. Right? And Casey is going to go, um, into a lot more detail and bring all of this to life with some of the real examples on our team.

But I wanna set the stage for why that impact mindset matters. So let’s start with the uncomfortable part. I think it’s, um, I think it’s important to start with why AI adoption stalls because it’s worth naming the problem before you talk about how we solved it Right, so AI’s been around for a little bit now, and even today, I’m talking to a lot of people, understanding how think- companies are thinking about this, and the conversation is obviously happening everywhere.

But when you start to dig in, and I’m sure some of you have seen this, when you start to dig in, the conversations often can seem still pretty surface level. There’s a lot of buzzwords. People are talking about a lot of tools, but I can’t really tell if people really understand and are clear on what is the problem they’re trying to solve. What are you actually solving for? What does success look like? And why is your current approach working or not working? And so I, I’ve really seen that the talk is there, but the operationalization of it, not so much.

All right. So let’s talk about why it fails in practice. It usually comes down to five things. First, there’s no clear why. Someone says, “Go use AI.” Everybody nods, but nobody really knows what that means, um, in their actual work. Second, there’s no ownership. When someone explicitly isn’t responsible for it, it becomes everyone’s job and nobody’s job. So there, if there isn’t a dedicated person or a team responsible for leading forward your AI adoption, it’s unlikely it’s going to move forward.

Three, no enablement or guardrails. So most people are not gonna be able to figure out what does good prompting look like or what are safe data practices so… or, or even tell them how AI is gonna help in their actual day-to-day work. So without examples or guardrails or guidelines and training, it feels risky and confusing, so people aren’t gonna pick it up.

Fourth, no accountability. If AI isn’t really built into your goals or performance conversations or the planning, if managers aren’t intentionally talking about AI fluency as part of your goals and as part of those performance conversations, people naturally start to prioritize things that they’re actually measured on, right?

Wouldn’t you do that? So it, it just ends up, AI just ends up being a side thing. And last, fifth, no governance. When there’s no clarity around data, privacy, what are the approved tools, legal and security teams tend to get nervous. That can slow things way down, and risk-averse teams just default to doing absolutely nothing.

So what’s the bottom line? Unstructured AI adoption defaults to noise, risk, or stagnation. So what does it look like when it’s done well? Naturally, it’s just the flip of everything I just named. So one, you tie AI to clear business problems. It’s not just a go use AI, but here’s the specific pain point we’re trying to solve.

You have defined ownership. You assign a team or a person to really be responsible for driving forward that AI adoption for your organization. Three, you invest in ongoing learning, and this is not a one-time training. I think the term ongoing is the key there. It’s not a one-time training. You build this continuous support because it takes time for people to build confidence, and that just takes time.

Four, you measure outcomes. Make sure it’s built into your goals, in planning, in how you evaluate whether it’s working. Make it part of the, part of the conversation so it cannot be ignored. And last, you let people know what is approved, what is safe, and where the guardrails exist. All right. So I’ve told you the ways in which it breaks and what you can do differently.

So now let me tell you about a place that figured it out, and I get to see-- say this because I’ve watched it happen. And this is what Zapier built, and I can tell you that it did not happen overnight. So when AI first hit the scene, right, Zapier was like most companies. There was curiosity, but also some hesitation, excitement, but uncertainty, nervousness, all of this mixed in together.

And the big shift really happened when we stopped hoping people would just go figure out what AI can do and started to create that structure around the learning and around the training, really setting that structure. And leadership made the real commitment. Our CEO paused projects and said, “Let’s refocus the business on AI adoption, refocus our people on AI adoption.” And Zapier really doubled down on helping people build the skill, build the confidence, and just have that sense of curiosity about what could be possible. And our Chief People and AI Transformation Officer, Brandon, who is here today, um, he often says, “It’s malpractice to not upskill your teams.” Gosh, and how true that rings today, right? I think about that a lot. And so instead of pushing for adoption, we focused on giving people space to practice, to experiment with real problems, to learn from other teams, to see what they were doing, to build their confidence really just one workflow at a time. And because the learning was hands-on, it wasn’t just theoretical, people started to see what AI could do for them, not just what it could do for the company. And AI slowly started to move from the edges into the center of the work.

So here’s a snapshot of tangible ways that AI can be baked into HR workflows, and this is how it is at Zapier. There is a lot on this slide. I wouldn’t focus too much on the, uh, reading all of that 'cause you can get access to that later. But the important part here is AI is built directly into our work processes. That’s really the bottom line here. It’s not something you do on the side. It’s embedded in how our teams plan, how they execute, how they make decisions. And from hiring to onboarding to performance management, it is just inherently now a part of our company culture.

And so I’ll, and I’ll be honest, right? When all of this initially rolled out, not everyone was thrilled. A lot of us, myself included, was like, “Okay, I have all of this free time, right? Now let’s add just that one additional thing on. My plate is so full. How do I add AI goals here?” But once people started building, once they started to see what it could do, that energy really shifted, and because it stopped being the homework that you do and started being something that actually made their days easier, made my day easier And I will say this too, when you think about AI transformation, m- a lot of companies think it’s a technology initiative.

And but when you really think about it, adoption, change management, any of those things, it’s really a people problem, right? It’s not a technology problem. And so who better to solve people problems than the people team? So that’s what we did at Zapier. The people team leads it, not engineering, but the people org.

And the way that we run it is in this hub-and-spoke model. At the center, we have the center of excellence as the hub. That’s the place where the strategy, the guardrails, the, the governance, the tooling standards, all of that is set at the center. And then we have pods that are embedded across every org in the company, right?

So marketing, support, talent, you name it. Every single one of them has an AI transformation manager who’s the DRI for the local adoption for that particular org or that team. And so the way that it’s set up is really cool. It’s centrally led, but it’s locally executed, and I think that’s the thing that really makes it stick because it becomes specific to what works for that org.

Here’s what it looks like in practice, right? So why does AI adoption work when it’s led from the people team? Think about all of the things that the people team owns. Hiring the right people, developing our existing employees, making sure all of it is tied to performance goals and expectations. It’s not just floating out there as a nice to have.

It’s really woven into how we evaluate and how we grow our people. Second, we define what AI fluency actually means. What does good look like? Like, where is that bar for what good looks like? Third, the culture piece. That’s where people team is what this sits with, right? We create the safe space. How do you create that safe space for people to experiment?

Um, how do we build that psychological safety where people can be okay to try new things but fail if, if it happens, and knowing that the right guardrails are in place. And fourth, it’s embedded into our job design and expectations. And so as we’ve matured as an AI transformation organization, the expectations that we have internally for ourselves has naturally become the expectations that we have for people coming in to level that playing field.

So what does it look like? That varies by role, right? What you expect from the people team is not gonna be what you expect from engineering. That looks different. So it’s very specific to the org, to the role, and directly maps to what each role does. But the baseline expectation across is consistent. You need to show curiosity, strategic thinking, and purposeful use of AI tools.

And the reason we do this for the people we’re bringing in, it’s because that’s the environment they’re stepping into. It’s only fair to set them up for success by evaluating if they have the skills for it. And AI is just part of how we work now. It’s not an add-on. And the key thing is, this is not aspirational.

This is what we hold ourselves to now internally, so naturally, this is how we evaluate people who are coming in. So that bar is very real All right. So how does this translate into work on the talent team in recruitment? The takeaway here is simple. We didn’t try to boil the ocean, right? We picked the areas where AI could create the most leverage, and we said no to everything else, at least for now.

We started by mapping our entire hiring funnel end to end. We picked a few high impact areas, so sourcing and screening, candidate experience, leveling our, our recruiters, our s- our talent team up. Those were the key things we picked, and every investment was evaluated against the three things at the bottom. Is it measurable and meaningful impact? Is it better for candidates, the fairness and experience piece? And can recruiters actually run this long-term, maintainability? And with that assessment in mind, we’ve kicked off four tiger teams in H one, um, which has been really exciting to watch. It’s, uh, these focused sprint teams that are driving real work forward.

So the- currently, the four that we have are AI-powered hiring plans, AI-powered sourcing, AI-powered scheduling, and AI interview prep. And the thing I’ll note here is that in every single one of these, be it the tiger teams, the upskilling sprints for our, for our own team, we again started with the problem and the impact, not the tool. Have I said that enough already? Um, the guardrails are also consistent across all of this. Humans are in the loop always. Hiring decisions are never made by AI. All right.

So in just a couple of minutes, I’ll be passing it off to Casey, who’s gonna lead you more into, like, the operationalization of what is the real impact of some of these big initiatives like the tiger teams that we kicked off. So she’s gonna talk through that. But before I do that, I wanna share something tangible from my own role. As the person who leads the AI enablement for the recruitment team at Zapier, I’ve had the chance to try a lot of things, which is great. Some have worked, some have not, uh, some have been surprising, and I’ve really learnt a ton along the way.

So what I’ve learnt the most is what it actually takes to move a team from talking about AI to actually building with AI every single day. So one example of what this looks like, we ran a skill sprint. It was one week. It was a hackathon week that we ran. Every single person on the TA team was tasked with building one skill using Claude. It was simple. It was time-bound, it was specific. Yeah. And this was not optional. Every single person on the team had to build it. You could be a manager, you could be an individual contributor, you had to build. And we gave them examples. Here’s the starting point, so you are not… If you’re, if you need some inspiration, here’s a starting point for you.

Here’s the structure we’re gonna follow as a team, and make sure that the work is shareable at the end because we can all learn from each other, because if it works for one person, it could very well work for another. And the reason for really pushing this is because you really need to build that muscle, right? Of building. So it was short, focused effort. You b- pick a problem, build a solution, share with us what was the before, what was the after, and what that impact was After we ran the hackathon week, the big thing we wanted to make sure was that we did a showcase of what everyone built. So we came together as a full group.

Everyone had to share what they built, um, the problem that it solved, what they learned along the way. So how did they iterate on, on the project, on the problem, on the build? What did they learn? We gave them a PowerPoint template. It was just a slide they had to use to share. Um, and if for any reason they missed the call, they had to share it on Slack, and so it was no exceptions.

And really, the main piece is it was twofold. Yes, the accountability. We said we would all build, so let’s all build, and you need to present. But also, the other piece was to really spotlight and share all of these cool things that people were building, and what came out of it was amazing. We had 13 skills that were created that are currently now all on Zapier for any Zapier customer to use as a template using our MCP server, and this was something we built in a one-week period. And it’s also a sense of pride for our recruiters, right? To have something and have it up there on a website, like, that’s amazing. So it was an amazing piece to, like, just start with a problem, the desired impact, and then stay human in how we make the whole thing work.

So that’s actually the perfect bridge for me to, um, for where Casey will take you next. She’s gonna show you exactly how this kind of thinking looks when we go into real case studies- What the things that we built and how our team did this. But before I hand it off to Casey, I’m gonna take a second to brag on her and get her just a little bit embarrassed, because she loves that. Um, I’ve worked with Casey for a long time, from well before Zapier, and I will say this: she makes me better, she makes our team better. The breadcrumbs of the work that she does is all over our team, and I hope at the end of this you will get as much from her as I have. And so with that, Case, I’ll hand it over to you.

Casey Firey: Hi, I’m Casey. I work in talent strategy and operations at Zapier with Anita and Brandon and Tracy and Dan, lots of us around here. And so I am so excited to take the next few minutes to share with you the impact that we’ve seen AI make at Zapier, and how we are specifically thinking about it in really tangible, concrete ways on the talent team. So before we dig in, I’m gonna paint you a picture first. So the first one is Boston. Anybody here actually from Boston? Whoo. Okay, lots of you. Keep me honest. So the first street is Boston, right? So that is one of the oldest cities in the country, and a lot of these streets started from just whatever made sense at the time. They started as walking paths or maybe trade routes or even, like, followed livestock trails, whatever just made sense whenever someone first walked it. Um, and as the city modernized, innovation just layered on top of what already existed. So dirt paths became cobblestone, cobblestone paths then became pavement. And I say that to say that I think our industry’s gut reaction has often been to just layer AI on top of existing workflows and make them faster, and it’s valuable.

Like, speed is great, but it is only one dimension of what AI can actually do for us, and it still leaves a lot to be desired if we’re only thinking about how can we make things faster. And if you have ever sat in Boston traffic, you can say amen. Um, and then the second city is New York. So New York took a very different approach, and before they paved any of their streets, they stepped back and they asked the bigger question, which was, “What should this system actually look like at scale if we were building it from scratch today?”

And we have that same opportunity with AI in that we can say, “Let’s just make our processes now much faster,” or we can step back and say, “If we are designing this today, what does an excellent system actually look like?” When we talk about AI and the impact it can make at Zapier, we organize everything into three categories.

So quality, efficiency, and experience. So quality. When we talk about quality impact, we’re really talking about raising the bar on the decisions that we make. With AI, this shows up as clearer signals, stronger documentation, more consistency across hiring teams. And quality is not about speeding things up, but it’s making sure that outcomes we’re producing are sharper and more aligned with what the business actually needs.

And you’ll see in some of the case studies we’re about to do some examples where AI did not make things faster, but it made them better. Efficiency. This is the one we are all super familiar, right? Efficient- efficiency looks like faster time to fill, fewer manual handoffs, decreased interviewer hours, fewer back and forth.

All the ways that AI can help all of our leaner teams do more with less busy work. And then experience. Experience for candidates, experience for hiring managers, experience for our own team and for recruiters. AI creates leverage here by making the process more transparent, more consistent, and honestly just, like, more fun for everyone.

We have better communication, stronger expectations, fewer surprises, better preparation, just a more supportive workflow for everyone involved And these three categories for how we think about impact give us a shared language for evaluating what’s worth building and keep us focused on the outcomes that matter most at the business level.

So bringing it back to Boston and New York, Boston only made the efficiency play. They laid pavement on top of a system of roads that needed more than just a new technology. New York paused, assessed their system, and then executed the plan. And so we’re gonna talk today about how to think about the impact that AI can make, and I want us to think about when we’re adding AI to a system, we need to step back and think about what do we actually need this system to achieve first, and then the impact that we need, whether it be quality, efficiency, experience, becomes super clear. These three impact categories, quality, efficiency, experience, do you know them yet? Um, actually ladder up really nicely to the hiring excellence framework that we’ve been talking about all day. So quality maps to Ashby’s evaluation rigor and strength of our hiring decisions. Efficiency aligns with operational excellence in the way we run our processes, and experience ties directly to how we attract, engage, and support candidates and hiring managers throughout the hiring process.

And so as we move to the- through the rest of this session, I just wanna keep that in mind and that these categories are not new. They’re just reinforcing the same pillars that we already use in defining hiring excellence, but just bringing AI into the picture as a force multiplier. And so we’ve discovered three distinct ways to think about AI’s impact.

And so now I’m gonna take you into the fun stuff, which is the nitty-gritty of what my day-to-day job is and projects where our team has really gotten to see that come to life So in the next few case studies, we are gonna talk through the pain point that we were experiencing as a team, the history of what we had tried, the desired impact, right, when we stepped back and said, “What does this actually need to look like?” Then the tool or solution we landed on, and the resulting outcome.

Okay, let’s jump in. So the first pain is recruiter bandwidth. And, like, listening to all of the sessions today, I realize that we are the only ones experiencing this. That was a joke. We all are experiencing it. Everyone’s busy. Um, but recruiter bandwidth, it just could not keep up with the hiring volume, right? And everything that came from making one single hire. Recruiters are spending huge amounts of time chasing hiring managers, and details, and doing repetitive tasks that just did not need to be manual.

And so the history of what we did is that we, like Anita said, we did a minute-by-minute process review, and it confirmed loudly what the team was feeling. And that was that the pain wasn’t one giant bottleneck at a certain part of our process, but it was dozens of micro moments throughout the funnel that were just stretching them so thin. And so the desired impact became super obvious at that point, which was we need a faster time to fill with a lean team, which is, like, everyone’s goal. Um, fewer manual handoffs, better recruiter experience, less chasing, less busy work. And then ideally, when all of that is true, we can pull recs forward and deliver more b- business- more outcome for the business than we could without

And then the tool. So after the impact was crystal clear of what we needed to be true, the right solution came into view, which was our AI-powered hiring material workflow that we built natively in our Zapier product. So before, like I said, recruiters were doing tons and tons of manual outreach to collect role details, pulling information from this framework and that framework to build all of those materials upfront from scratch.

And the kickoff of a role was where a really heavy concentration of those paper cuts lived. Now, the entire beginning of the process is orchestrated through automation and AI. So once a headcount is approved, it’s automatically added to ChartHop. From there, a chatbot automatically reaches out to the hiring manager to collect structured inputs.

Then AI uses those inputs in our internal frameworks and repositories to create a job description, application questions, and an interview plan all in one run. From there, a Slack channel is automatically created, and the materials are shared instantly. Everything is completely orchestrated across all of our tools and ready to launch without recruiters having to run a million miles an hour to get all of that ready, which is a real bummer because that’s their favorite part of their job.

Um, and so the tool wasn’t the starting point, right? We didn’t just like wake up one morning and think like, “We’re gonna build an AI-powered hiring plan.” But it was the natural output of when we realized, okay, this is a problem. This is what we need this to look like, and then we figured out like, okay, this is what we need to build And then the outcome when we piloted it is that the transformation was super real.

We saw a faster time to fill. We have a lean team. There were fewer manual handoffs, and the recruiter experience improved because the system was doing so much more of the heavy lifting than it was before. Okay, next case study. So a pain we experienced is that references have become a huge part of our hiring process.

And as the emphasis that we have put on that part of our process has become more and more, a gap became really obvious in our system. And that is that written references were not giving us the detailed insights that we needed, and live references, while a goldmine of information, were just really, like, operationally clunky and kind of slowed our momentum at the end of the funnel when, like, everyone really wants to go fast. And so we needed deeper signal without slowing the process. The history of what we had tried. So we were running both written and live references separately, and neither system was giving us what we needed in a scalable way, right? Written references, so quick to send out, dime a dozen, quick for referrers to s- to send back.

But you know what? They’re generic, and they don’t really close any, like, candidate-specific evidence gaps. And then live references, like I said, so good. Such good insight. But coordinating them was so clunky. It was time-intensive for recruiters, time-intensive for hiring managers. The candidate experience wasn’t great.

And so the desired impact, it’s probably obvious to all of you at this point, was, like, we need high decision confidence at that point in the funnel. We need a smoother process. We think this is gonna elevate our quality-of-hire outcomes, and we needed a way essentially to generate specific role-aligned questions for the candidate and get that information back quickly

And then what we built from that became super obvious, and we built this again natively in Zapier. So the first part of this system, what it does, it’s AI analyzes everything that we already know about the candidate. It brings together and orchestrates interview scorecards, Bright Hire transcripts, the job desquish- description, our quality of hire traits, the evidence gaps that were flagged during the interview process.

So now, references aren’t happening in this vacuum at the end of the process, but AI understands the full context of the candidate, the full context of Zapier, and the full context of the role before generating anything. Two, it generates custom reference questions tailored to that candidate. So based on that analysis that it does, AI writes individualized questions that targets the exact areas where we still need more clarity to be confident about hiring this candidate.

Hiring managers then, which you can see in this image, get the opportunity to approve, say, “Yes, this looks perfect, send it out,” or they can edit it directly in Slack and say, “Actually, I’d rather get more context about ABC.” Three, then it’s gonna fully orchestrate the live reference workflow. So again, huge pain point was just, like, getting those things on the calendar.

And so now the system will automatically send a scheduling link to the references that were submitted, follows up if they haven’t booked, and nudges the hiring manager again in Slack, which is where we live, to confirm whether or not the call was scheduled, all without recruiters having to lift a finger.

Then finally, it’s gonna prepare the hiring manager for that call. So it’s gonna send another message to the hiring manager saying, “Hey, just a reminder, here is what we need more information on to make sure this candidate is just really an excellent addition for this team.” So the hiring managers walk into that call knowing exactly what they need to walk out with.

And then from there, it synthesizes everything. So the- once the written references come in and the live references are done, AI produces a clean summary that highlights what was confirmed by the references, what’s new that the references told us, did anything contradict something else that we had already seen, any remaining risk areas or gaps, and then what information did we get that filled the gaps that we identified earlier in the process?

And so hiring managers then get one high-signal, decision-ready view that makes them confident as they’re entering that last stage of our process And then the outcome of this tool has been really exciting too, in that it has increased reference quality. We’ve gotten positive feedback, unsolicited, from references saying that our process was great.

Um, hiring managers are loving the increased signal that they’re getting, and we are saving buckets of money, which no one is mad about.

Okay, so you have seen how we are mapping the impact that we need from AI at Zapier. Now I’m gonna slow down and I’m ma- gonna make this super practical for you. So we’re gonna walk through how to apply this same thinking, which is leading with impact first, to the right tool and process so that that becomes super clear as you’re thinking about how to do this in your talent teams.

So if there was one thing I would want you to take away from this session, it would be that this approach about thinking about AI’s impact before selecting the tool or the solution. And if you were to sit down at the happy hour that’s right after this, after Kenny’s session, um, if you were to sit down and map this out on the back of a napkin, this is how I would do it.

So I’m gonna walk you through one last Zapier example of thinking through this framework. So the pain in this one is that we saw a new pain at the top of the funnel, and that is that application volume spiked, AI-generated resumes became indistinguishable from each other and from the real ones, um, and some candidates, believe it or not, no one in this room knows it, are submitting fraudulent materials.

And so every application looks polished, everyone seemed perfectly aligned on paper like they were born for that role, but none of it told us what the candidate could actually do in the day-to-day. Recruiters were spending huge amounts of time trying to separate the signal from the noise, and that ear- those early stages became a huge bottleneck that we just were not scaling as a team Historically, we relied on recruiter screens and application questions to surface real signal early in the funnel.

Things like motivation, depth of experience, alignment with the actual responsibilities of the role. But with those new increases, resumes became less reliable as a filtering mechanism, and the model just became unsustainable. So the desired impact, again, becomes super clear in that we needed a reliable, structured, early funnel signal that resumes were just no longer providing.

We needed a way to validate experience directly from the candidate, not from their AI-polished materials. We needed reduced recruiter time spent on repetitive screens with candidates who weren’t qualified, and then we needed higher consistency in the way all of that was gathered. And of course, we still needed humans making the decision And so from there, we decided to test an agentic screening tool at the top of the funnel.

We piloted it with a few specific roles, and then a working group of TA leaders and folks from that tool would meet weekly to discuss successes, hiccups, everything we needed to talk about so we could make real-time adjustments as we were seeing if this could solve our problem. And then that pilot delivered exactly the lift that we were hoping for.

The agent ran a structured screen, recruiters received clear summaries covering alignment, depth of experience, motivations, and we immediately reduced the noise from those AI-polished and AI-generated resumes. And the team got meaningful time back, spending less time on that finding a needle in a haystack and more time on deeper evaluation of great candidates that we’re excited to get to know.

Okay, so we’ve talked about a few case studies that are so fun for me, and as someone in operations, I am in the weeds, and I am working with tools. I understand what’s possible. Like, maybe I am the first person to get to experiment with something. And so it is so easy for me to look into my crystal ball when I see a solution and just get pumped about it.

However, if anyone else is in operations, you know that you have to bring other people along with you. And so we have to bring along recruiters. You have to bring along hiring managers, candidates, company leaders, other stakeholders. And so I just wanted to throw your way a few ideas on how to create that buy-in with any stakeholder that you’re working with that may not experience that same like love at first sight that you do when you see the impact that AI could make.

So when I’m putting together a memo or an impact story for stakeholders, I keep the structure super straightforward. I want everyone to see the problem, the meaning behind it, and the path forward, all in one place. So first, I start with the problem. I lay out the problem from every angle, qualitative, quantitative, operational.

Here’s what’s happening, here’s how we know, and here is how big this impact is. And the goal for this section is just clarity. We all need a shared understanding of the pain before we move into any solution. Two, then I synthesize what the data is actually telling us. I don’t just give a million different data points and expect them to connect the dots, but I connect the dots for them. X problem is costing us or causing us Y, which is leading to Z. So I turn all of the raw information into the clear narrative so that I know that everyone is seeing the broader implications of what we’re talking about. Three, I share a clear path forward. Based on what I’m seeing, here is what I recommend we do.

I give them a well-reasoned direction that’s rooted in the insights that we’ve already agreed upon in steps one and two Then four, I go to the desired outcome. So here is what I think you’re gonna get when this is awesome. So this might cost us ABC or the trade-off you might see might be XYZ, but here is the value that we anticipate, whether it be savings, lift, quality gains, reduced risks, you name it.

Um, but it helps us frame the investment or the change in practical terms and then connects it back to the business goals. And then five, finally, what I’m watching. I talk about the guardrails. Here are the metrics I’m monitoring. Here are the thresholds that matter. Here’s what would prompt an update. I want stakeholders to know that, like, I’m not just, like, throwing this out there and assuming it’s gonna work, but I’m gonna keep an eye on it. I’m gonna update you if anything changes. I’m gonna get your feedback. You’re gonna hear my feedback. It lets everyone just kind of, like, rest easier knowing that we’re not just, like, setting it to sail and letting it loose. Um, and then also, I cannot leave this slide without giving Brandon, our A- our chief people and AI transformation officer, um, another huge shout-out in this for his leadership and empowerment of his team to really, like, have these big ideas and dream big and really, like, redesign our roads to get us to wherever we wanna go.

So as we wrap up, I also feel like it is important to say that, like, this is so fun to talk about. Like, you could have let me do like 20 case studies, and I would have loved it. Uh, so come find me later. But I also wanna say, like, it’s not rocket science. Um, at the end of the day, this is just really structured change management applied to a technology that’s just changing really, really, really fast. And yes, that speed can make it feel harder to kinda like get your arms around it or know exactly what to do, but if we keep impact, those metrics that we’ve all been watching for years and years and years, in the crosshairs when we’re making the decisions and let those be our North Star, then we’ll still get the adoption and the buy-in and the outcomes that we’re looking for even as AI keeps evolving really, really, really fast.

So thank you all so much.