The CFO’s AI Playbook: How to Roll Out AI Across Your Company, with Ryan Roccon (CFO at Zapier)

As models got better and products were released, our attitude has always been, and as the cfo, I have to back this wholeheartedly, has been move forward aggressively. Like, don't let cost be the burden. That sort of stops you from experimenting and learning and growing and finding what works. AI is moving from experimentation to expectation. But most companies are still figuring out how to turn scattered use cases into real business impact. Today's guest is Ryan Rakan, CFO of Zapier. Having progressed from controller to cfo, Ryan brings a practical finance lens to one of the biggest questions facing companies today. How do you fund AI aggressively, give teams room to experiment and still keep the right guard rails in place? At Zapier, AI isn't a side project, it's a top down mandate, part of how the company operates and now a key part of how they hire. Every team has broad access to AI tools. And since 2022, Zapier has been redesigning the way work gets done. In this episode, Ryan shares how Zapier is approaching AI adoption at scale, why curiosity is rarely punished, and why the rise of AI puts a new focus on the CFO as a capital allocator, not a gatekeeper. We also get into what finance leaders should track, how to balance speed with discipline, and how to move AI from experiments to real adoption. This is the gross minded cfo. Let's dive in. Hello, welcome to another episode of the Growth Minded cfo. I'm Lauren Pearl, here with my wonderful co host Alex Luisi. And today we are joined by Ryan Rakan, CFO of Zapier. Thank you so much for coming on the show, Ryan. We're so excited to have you. Yeah, thanks for having me. So, as we kick off, maybe you can start us by just telling folks a little bit more about who you are and what you do and then we'll get started. Yeah, yeah. So as mentioned, I'm the CFO here at Zapier. Zapier is a workflow automation tool. We ultimately present AI automation in a governed way. And I've been with the company for more than seven years now. Started out very early stage and have since sort of expanded my role as cfo, managing all of the standard CFO functions, finance, accounting, tax, but also managing other interesting functions like data ecosystems in channel biz ops and a whole bunch of other fun areas like pricing and packaging. So it's been a really exciting journey with a company that's been very exciting as well. So can you tell us, just before we dive into like exciting AI topics, can you tell us a Little more about more your career at Zapier around when you joined, like what was your position and did you want to be the CFO in the first place or did it happen to you to be like, what is your journey and what was your intention about the journey? Oh boy, we could use our whole 45 minutes. So when I joined, I actually joined as a controller and there was only, we were about 150 people and we had a CFO at the time and a couple of accountants. And right off the bat my mandate was like many sort of new hires within the finance functions, especially the controller, was to just build everything. So, you know, the FP and a team. Finance, accounting, tax, indirect tax, procurement, payroll, legal, contract negotiation, contract tools, all of those things. None of them really existed when I joined Zapier. So I spent the first few years building out the team quickly sort of went up the corporate ladder, let's say, and then found myself with the opportunity to step into the CFO role a few years ago. And so now I've been in that seat for close to three years. Wow, that's a big progression from controller to cfo. I like your ethos of building things. Was there something in particular that you built that really you think was part of that journey of going from controller to now more of a strategic player? What would you say is sort of your foothold that got you through that progression? Yeah. Oh, I had a really interesting sort of career defining moment in many ways a couple of years ago. So before that CFO transition. So Zapier is a company obviously for those familiar product led growth, self serve, sort of business freemium model. So we've achieved the growth that we have because of that sort of product led growth, flywheel type model. And over the years we've realized that that only scales so far. You really have to sort of step into this sales led growth model and this upmarket expansion. And it was something that we had been pushing with the company for quite a while. But to be frank, there was a bit of an organ rejection kind of happening because we had such roots in this very deep sort of customer obsession with like this very large customer base. And this idea of like a sales culture felt a little, a little strange for us. And so I spent quite a few months really digging into the data and mapping out the entire history of Zapier and all of the stages of growth and the inflection points and what changed to get us to the next stage. And I modeled that out for the next really to the end of the Decade at that point it had been, you know, seven or eight years of sort of like projections and I started to sort of show where gravity would play its part on a self serve growth basis and how you have to start to stack S curves and sort of when those things become not just important but required in a sense. And I package that up into again after doing a lot of very, very deep modeling I packaged that up into what ended up being something like a 50 page. And I know shame on me, who wants to read 50 pages? But the reality was like a 50 page sort of memoir that I shared with the entire company. And ultimately it was this is our path to, to the end of the decade to these growth targets that we've set. And this is why this move up market is not just a nice to have but a need to have and really spelled it out to the whole company. And it became sort of this rallying cry where everybody in the organization could answer like this is what we're doing, this is why we're doing it. This is the timeline we're doing it on. And there's no sort of disputing it because the math is the math and we all agree that this is the, this is the path forward. Wow, that's so bold. That's a really incredible thing to go and do and kind of like put out to the entire organization. We had another guest on the show, Ryan, who shared the same story around like she wrote a memo on the company when she joined. And I think what you're telling here is I think super interesting for everyone listening this episode and who wants to become like I don't know if you want to become a cfo, but kind of like probably like going up into your carrier is probably like putting yourself in shoes that are just probably like of something that is not like on your job description right now. I don't think that you were asked to produce that document. And just being ahead of that, that's a great learning. Do you have any tips for people when they kind of like navigate the scour changes and sometimes they were afraid of doing such moves? Right. You feel that maybe some people might say this is not your position, you shouldn't be talking about what is going to happen in the next Dec for Z. How do you navigate that for new joiners or maybe like earlier people like in the carrier? Yeah, it's a good question. I think there's probably a few things to keep in mind. One is that curiosity is rarely punished. And so I think that's something that I Look for in every person that I hire is a very deep sort of curiosity mindset. And this willingness to, when you see a thread, not to just let it sit there, but to pull on it and sort of go deeper and, and share a perspective. And then from there, I think the other thing that folks can do is qualify their statement. They can say, hey, I might be wrong, but I'm not silent. Sort of. It's better to sort of put it out there and have that disagreement than it is to sort of sit on it and not know whether folks feel the same way. And so I ultimately will always advocate for that. You know, you can put that out there as a disclaimer. Like, I'm not entirely sure about this. This is a theory, it's a thesis dispute it. Let's, let's debate. But, but again, silence is sort of disagreement. And so that's the, that's the, the mandate that I would share. Yeah, that's a great point. So if I like. The one thing that I like a lot, and this is the same for us in the company, is around making sure that you speak out, but just creating the condition of like, your message is going to be well received rather than just creating this kind of like tension or so I think it's a, it's a, it's a great way of like pushing those ideas forward. And it's also good to pick a company where this is the culture. That's also the point. Right. Being in the right place. Yeah. Because sometimes, you know, I see so many people that just join a company and they don't just realize after the fact that this is a company where you're not allowed to just make the statement or. And I think this is. I fully agree with you. This is probably like one of the most powerful things of how an organization is evolving, like letting people being able to have this level of ownership. But it's not always the case. It's not always. But it sounds like it is in your organization. Actually, it's funny, as you're telling this story, I know that you have a lot of thoughts on AI and AI adoption and finance teams. And I think we're going to end up talking a lot about that this episode. But what strikes me about your story is it makes sense that bats stuff that you're exploring, given the company that you're coming from, the way that you've been sort of celebrated by having a strong opinion and pushing the company forward. And also sort of your attitude about curiosity and sort of being willing to state an opinion and what you think, even if you're not completely sure that it's going to be right. But having the conversation, maybe we start talking about that same early days experience with some of the AI tech and exploring it on your finance team. Can you pull us sort of back into the room of when that first started coming into the conversation and what that was like and how you started deciding that this was worth exploring on your team? Yeah, I mean, so specifically within. Within a finance lens or just in general? That's a great question back because I think it kind of depends on where it came from because I guess I'm curious where it started. Yeah, no, that's, I think that that sort of reframing is important because at Zapier, the thing that I'll, I'll, I'll share about our company is, I'm sure a lot of folks say this, but Zapier is one of the most AI forward and sort of forward thinking companies that I've ever been around. And we called an AI code red in 2020, I think, or 2023, I can't recall, and started to redesign the company around it back then because we saw the writing on the wall, we saw what was coming and so, and from that we started to, as models got better and products were released, our attitude has always been, and as the cfo, I have to back this wholeheartedly, has been move forward aggressively. Like, don't let cost be the burden. That sort of stops you from experimenting and learning and growing and finding what. And so in many ways there was a lot of discussions early on around, hey, we want to build this tool, but it's going to be expensive or we want to buy this tool and it's going to be expensive, should we do it? We don't know the ROI on this yet. And the reality was when we sort of put that math and looked at that math and sort of put it on paper, ultimately a lot of things were based on assumptions that we just didn't understand. Right. But we still said we're going to go do this because the cost of not doing it is so great. The pace is moving so quickly. So we've been very sort of adamant about getting our hands on the tools and making broad access available from the early, early days. And that extends obviously into the finance work. But I'm happy to sort of explain more where it's useful. Yeah, maybe we can get a little bit into how it's evolved to kind of set context on where it is today. That could be really interesting. Context. Yeah. And you just mentioned broad access. I'm very curious about that as well, because it seems that you have a contrarian view on this, and I am very much aligned with you. So I'm really interested about hearing your thoughts on this. Yeah. Okay, so let's talk about access, and then you can tell me if you have where we want to take it from there. So, yeah, I mean, the very simple version of this is no matter what role you are at Zapier, you have access to almost any AI tool that you can imagine, like point blank. So we've got every anthropic tool, all the OpenAI tools. We've got every single person on Zapier using cursor. Every person you know, you're an accountant, you're on cursor. Everybody's got access if they want it to. Meeting recorders like Granola and Fathom and all these other great tools. The list goes on. And then there's specialty tools in cases as well. So that is sort of the first step, which is ultimately everybody has access to these things. And as you might imagine, it's quite expensive. Yeah, we see the belt. The bills rack up pretty quickly. It also comes with risks. You know, you hear about stuff getting shared or leaked or whatever the case is. And. And we've had situations where internally folks are sharing a doc that shouldn't have gotten shared with some team because it had private information in it. And it's sort of one of those things where it's a risk, but that risk is worth it in many ways because of the speed of experimentation that we get as a result of that. We actually have a formal AI transformation program at Zapier, and that is led by our chief product technology. I'm sorry, our chief people and AI Transformation officer. His name is Brandon. And then we also have named functional leaders who are the AI transformation leaders for every function. So I have one in finance as an example, one in biz ops, one in product and engineering, and so on. Oh, interesting. And their job is to ultimately look at the tool usage across their teams and make sure and help those teams prioritize big shifts in sort of capabilities that result in things that we can clearly measure and outcomes that we can point to. And they actually end up publishing a quarterly case study by function to say, like, hey, here's what we built and the capability that we've unlocked. So it's a few examples of how it's evolved, but happy to dig in wherever useful. No, but that's a really good point, because I Think you mentioned everyone has access, which is not the norm. That's like quite rare. But then the question once you have access is, well, how people actually using it. Right. For us, typically in the company, like we do like literally weekly training where everyone is kind of like coming up and showing like this is what I've done with Claude. Right. And it's, and it's shocking to me to see like how people are still learning six months later, a year later, because some of them are just going to take it and just run with it and try to build and break everything and some are going to do the opposite and just keep on doing the old way and stuff. So I was kind of actually curious about Adzep here. You mentioned some of the programs, but how do you actually not necessarily force, but how do you ensure that this is not I have access to it, but I'm actually using it, testing it and making me more productive. Yeah. So there's a bunch of different vectors in which we do this. So I'll talk about each other. So the first is the fact that we have made it effectively a top down mandate. We've said this is part of your job, you have to use these tools. We talk about a lot about this publicly, but when we hire, that is part of the rubric. We are very clear about our expectations on how competent and capable you are with AI tools and your understanding and willingness and experience having used them. And so internally it's the same, the same as being applied for the folks on the team. Like we have an expectation. You can't sort of stand by and just watch these things happen. So that's the first thing, it's a mandate. The second thing is that we do have a lot of telemetry. So as part of that we try to make sure that we can actually see where the pockets of usage are happening and where there are gaps and work with managers to say, hey, it seems like we're not doing enough over here. Can you work with your team to find out why that is, what's blocking them? Is it a lack of willingness? Is it a lack of access? Is it because they're too busy or try to get to the bottom, the root cause of why they're not using those tools? And then to your point around this idea of some folks just running away with it, we make sure that those types of use cases and those capabilities being built are broadcast very loud and clear. We give them a platform, we do hangouts, we do all these sharing sessions. We've got dedicated space for those folks to go and encourage other folks to see what they're building. And the last thing we do is one of the biggest blockers that you'll hear from those who are maybe a little bit reluctant to go and build is, I just don't have time. And so we make the time. We give folks a week, a quarter where we say, clear your deck. No meetings, no deliverables. This week only is go build, go build. And at the end of the week, we have prizes. We've got big sort of like showcases. We try to make sure that the folks who are doing this and doing it well get celebrated. And so we create all the mechanisms that are necessary to sort of have these folks building and, and sharing that work with others. That's awesome. I love these examples of things you can do as an organization to push and encourage more AI adoption, more usage, more use cases. It strikes me though, that you had that code red years ago and that's the stance now. So how as an organization did you get to that point? Was it sort of like one day it was just all in, everyone has access, like, let's turn the guardrails off. Was it a gradual change? What was that like? And did you learn anything from how you went through that progression that you might change next time if you were starting new with a new org? Yeah, I would say it was reasonably gradual. The reason why I sort of pause and say it that way is because ultimately, in many ways, it was dependent on the evolution and the capabilities of the tools. As you saw tools getting launched and the capabilities sort of expanding so quickly, that's when you sort of go, oh, wait a second. We just unlocked a new layer of capabilities and productivity. Like, we have to get on this. And so we went as fast as we could, but it was gradual in many ways because it followed the curve of sort of capabilities of the models and the model providers. With that being said, as soon as we saw the opportunity, we didn't hesitate. It wasn't like, oh, Kershaw, this is interesting. It was like, wow, this team is building this second brain idea where their entire team's context is living in this shared space and everybody can prompt against it and use it and understand where projects are and all these things that has to be company wide. We have to make that happen now. We're not going to wait. And so it was gradual, but sort of gradual with big spikes of adoption and, and sort of a push across the org. Super cool. You just mentioned a lot of actually broad names around, like, you know, the Tools that everyone is using. And I think this is a statement that you made while we were preparing that episode around, you know, broad tools versus, like verticalized tools. Like, what is your take on that in terms of why. Why do you want to push for more like using broad tools rather than verticalized one? It's a good question. Yeah. And I think obviously I'm a bit biased because Zapier plays a large part of our ability to do these things well in many ways. When you look at some of the AI products that exist today, some of them are traditional SaaS, products that they've layered on some sort of, you know, oftentimes a wrapper over some, some model, whether it's Clotter or chatgpt or whatever the case. And what we have found oftentimes is that when you get to those types of vertical tools, you can achieve the same outcomes that they're. They're sort of selling and oftentimes at very exp price with a tool like Zapier. The power of Zapier is the ability to connect across 9,000 applications and then layer in both determinism. So triggers based on an event while you're sleeping, but then ultimately AI as a step. So you can sort of do, if you're in a procurement tool and you're trying to route based on priority or owner or dollar amounts or whatever, you could buy the sort of vertical AI routing system from that tool, again, usually quite expensively, or you could take the integration that's built on Zapier and build an automation that says use this AI step to route this thing based on these conditions. And so I find a flexibility that comes with being able to connect across all of these systems, not get stuck into sort of walled gardens of these individual ecosystems is really powerful and quite cost effective. Yeah, very. You've talked in the past about not being too focused too early on roi, but measuring everything. The thing that makes me curious about is one sort of if you're measuring everything, if not roi, what are you looking for? What are the things that you're looking for in that measurement? And how are you then making decisions based off of that? Yeah, I mean, ultimately there's sort of an order of operations here. Okay. The first thing is you have to have the access has to exist. Right. You have to be broad based access, I think is my opinion. The second piece is in order to measure outcomes, you do have to build the telemetry. And so for us, we spent a lot of time being very explicit around all of those sort of metrics that we want to pay attention to and going deep to understand eventing because ultimately you can't understand usage and be able to say what outcome are we sort of impacting without. Without that. And then third is where you start to actually optimize and look for efficiency gains and sort of increase usage or draw it down depending on how efficient or effective it is. I think you have to go in that order. And the reality is that early on a lot of folks will find that it's difficult to put your finger on what the ROI of something is. And I think if you get obsessed with that too early, that you might find yourself sort of stuck in this analysis paralysis and not learning quickly enough to understand what the tools are capable of. And so for us, I think that the major sort of push is build the eventing, take the time that goes into that, but also recognize that you shouldn't make sort of knee jerk reactions based on the information you see on day one or even day 60 because the capabilities are moving so quickly. You really need sort of like a longer view to understand how impactful these things are. Yeah, and, and I think I fully agree with you, Ryan. I think the, as a CEO, this is something that in the company I'm always telling the team around, I'd rather look at a dashboard that is wrong than not having this dashboard in front of my eyes. Like, this is how I push it. Because oftentimes like I get this pushback from the team, they say, oh no, like we can't, we can't use that dashboard because the data, the underlying data is not correct. And I'm saying like, yep, let's first like put it up and then we'll fix it over time because that's by looking at it every day that we're going to fix it. And I also like a lot what you said around the roi, because if we too much focused on the ROI first, like you might just do the wrong decision, make the wrong decisions as well, you know, because you might be optimizing for the cost when you should be actually focusing on the outputs. And that's actually, I think when we circle back to this idea of the adoption curve and the time needed to actually get used to it, I think the short term error, it might be negative because when you say you're going to take a week off to just learn on how to use glute code or this probably might be seen as a waste of time in the first place when it's probably going to yield huge error in the long term. And I think that's a big challenge. What I think is really interesting, and that's where I would like to get your view on, is that very often CFOs do not think like this. They are more on the brake type of things than on the accelerator. How did you become that cfo? That's a real question. Because we see a lot of growth minded CFO that are similar to you. But we know that this is not how Most of the CFOs are out there. They're sometimes more on the. We need to be cautious, we need to be careful about costs. We don't want to use all those tools that might not have an erra. Like how did you become that cfo? That's a good question. Yeah. Like an origin story, right? It is. I think in this case it was forced upon most of us, myself included. I don't think there's been a shift in technology that's so profound. Yeah. In any of our lifetimes. And I think to get stuck, stuck in a cost oriented mindset could be the biggest mistake of a CFO's career right now. Honestly, it's truly transformative what's happening right now. And for any of us to be able to say that we can accurately predict what the cost structure will look like. I mean, look at the model cost when they first came out versus today. There are fractions, fractions of what they were. And so if you made a decision on day one two years ago, you know, when some of these models released and said, well, we're not going to use it because it's too expensive and didn't revisit that on a daily basis, you would be sitting there way behind the curve as a result. And so I don't think it was necessarily that I have some predisposition because of, you know, something I've learned along the way. I just, I look at this as, this is a, a moment that if you don't make this sort of change and recognize that things are moving so quickly that you're going to be going to be left behind. Yeah. Did you get some horror stories of this mindset? Because sometimes like, you know, in the, in the long, like in the grand scheme of things, like that's the right mindset that we want to have. But sometimes, you know, like you have some moments where typically like the cost of one tool just go through the roof or is there like. Have you had any heart attacks along the way? I think is what Alex is asking. Tons of them. Oh yeah, absolutely. Absolutely. Here's the good news now, the good news is that cost as a heart attack is super easy to fix. You can just shut it down. Sure. Like top line growth, not accelerating, not sort of being at the front edge of your industry, not sort of pushing the boundaries on your product and what you're delivering to customers. That's a much harder thing to sort of turn the off switch on or turn the on switch on. You need a 50 page memo. That is so true. That is at the crux of this. It's like cost is easy, it's easy. But yes, I've had plenty of heart attacks. So yeah, one interesting example, this is a good one because this was powering our internal copilot feature. So copilot within zapier, like a lot of other copilots, when you log in, the very first thing you see is this sort of blank sort of input box where you can natural language prompt what you want to build. I want to build this tool, this automation, et cetera and it turns it into a zapier animation full of agentic capabilities. All that good stuff and a lot of technical sort of components here. But the tldr, at least how it was explained to me was we were stuffing a bunch of history into the context window for the prompt as users were returning and we were storing that sort of indefinitely and it was causing tens of thousands of tokens to be cached every time somebody came into this, this copilot. And what ended up happening was within a few weeks of us launching. And we didn't see it right away because it sort of took time to build and it took a few outlier users to really spike this. We had a multi hundred thousand dollar bill in a matter of weeks when we were expecting it to be in the like single digit thousands. Oh no, just right away we have one customer cost us something like $20,000 in a matter of hours. And the funny thing about it is everybody was, all the people on the team were like, oh my gosh, Ryan's going to just like rain fire down on us. What a mistake. This is awful. And it was like, no, that's fine, it's good. People are using it, obviously. This is great. We love to see the adoption. Can we go fix it? Can we reduce the cache window? Can we? Or the cache and the context window. Can we make this a little bit more efficient? And then immediately they said yep, we understand the problem. They went and fixed it. It was a matter of a day or two and we're good to go. But I like those problems I'd rather see a runaway cost associated with runaway usage than not see it at all because we're being so sort of cost conscious. Yeah. Did you use that example also as a way internally to showcase the mentality of not slowing down innovation in some ways? Yeah. We have a company, mbr and I remember this came up as one of the topics. This, this runaway costs. We're going to fix it. It's going to take so much time. And I use that as a bit of a platform to say I want everybody here to know this is not a mistake. This is a learning. This is, this is an opportunity to make it better. Like these types of, these types of events are not only expected but sort of demanded of us if we want to move fast. So yeah, I try to reinforce that and make it clear there's no mistakes here. It's just learning. Also, there's different. This is a great example of a learning experience and I love your call out of, you know, cost isn't the worst heart attack. It's really when you, when you have top line issues that it, that it happens. But there's different types of things that can go wrong. So I'm curious if you had any, any sort of activities you've done to mitigate risks sort of beyond cost, so coordination with IT or CISO to think about security concerns or data loss or data leakage. Are there steps that you have taken to ensure that when you do have those mistakes they're not catastrophic? Yeah, absolutely. We have a whole lot of guardrails. Yeah, it sounds like wild west over here, but we've got a ton of guardrails. When it comes to anything that touches customer data or financial data, those things are sacred cows. Those are going to be protected and have all sorts of guardrails around. If you try to sort of publish it on something that's public facing or if you try to put it in a repo over here, it's going to get blocked. So it has managed and built those things and we've put those guardrails in place to make sure that that kind of information stays very, very secure within the sort of internal productivity use case where folks have access to all these tools and they could as an individual run up really expensive bills. I again, I would kind of frame it in the same way that I just framed it for the, for the one that powered a customer experience. We love it when we see somebody rack up a 20k bill. Generally speaking, we love it. Let me caveat that when we see somebody rack up a 20k bill in a matter of a week, we're immediately like, what are you building? Show us. This is. And nine times out of ten, it's very cool. Whoa. We have never seen anything like this. This is super exciting. Wow. You're thinking about Zapier in a whole new way. People come with the most incredible ideas, and I've seen two, three, four of those, maybe more turn into actual big product features at Zapier or entire standalone products because somebody had some brilliant sort of harebrained idea and started coding, vibing, and built something. Every once in a while you get one where you're like, somebody built something that's almost like a loop and it's wacky and it's the highest model and you're like, okay, what's that? But that is the very rare exception. And again, I'd rather have a handful of those every so often while still having the sort of stroke of genius that comes with more of the uncapped usage approach. Ryan, I think we're going to take just a short break after that, but I loved everything you said around those examples because it defines the culture of the company. And I think that's probably the most important thing is not the one mistake or the one cool thing that was built. It's also how you use it as a platform to broadcast to everyone in the business that this is how you want things to happen in the business, as everyone seems to be aligned around this. And I think as a CFO for any listener, I think I often find myself guilty of just managing this as a one off and not broadcasting it enough so that people know that that's the culture. And I think that's a great learning. It's that culture of curiosity that you've already described in your very first pivot. Yeah, that's right. 1. Just as a tip for folks out there, one way that we. So I mentioned a few ways that we broadcast these things, but one way that really gains a lot of attention and traction is our CEO. Every Friday, he publishes a Friday 5 where he goes and gathers sort of the five most important things from the week that he wants the company to know. And in that Friday 5, almost always, most weeks, you'll see something that somebody built that's changing the mind of him or the exec team that's really pushing us forward. And so that kind of visibility really gets folks going. Like, I want to be on that list. I want to build something better. Super cool. That's super cool. Thanks so much for sharing this break. And then we'll come back and talk about more awesome curiosity culture insights. Thank you so much, Ryan. Still to come in this episode, many times CFOs, they often are represented and act as the gatekeepers. They're the know, the know people. Uh, and I think especially in this new AI sort of era, this is also a very important part of the job. But I think there's a reframing here where we have to be the capital allocators, not just the gatekeepers. Enjoying today's episode, don't forget to subscribe on your favorite podcast app and tell a friend about the show. Now back to the episode and welcome back to the growth minded cfo. We're back here with Ryan and Ryan, we've talked so much about the AI innovation at Zapier in this wonderful culture of curiosity. We almost got there before talking a little bit about how you adopted this AI forward mindset, but we didn't quite get to what you would Recommend for other CFOs, especially folks who are sort of stepping into that CFO role, who are trying to build that culture in the organization. So what tips do you have for them coming in and creating something similar to what you've been able to achieve? Sure, yeah. So I think the first step and something that's worked very well for Zapier is to create a center of excellence with clear ownership and ideally a single DRI over the whole program. So as I mentioned, we have our AI Transformation officer who's responsible for the whole program and then we cascade that down to those functional AI leaders that that sort of top down mandate, clear single threaded ownership, measurement of impacts, deliverables. Like I mentioned, we have quarterly case studies. We've got sort of this backlog of work that each team is focused on. I think without that center of excellent excellence, you will have a bunch of folks chasing a bunch of disparate ideas with different sort of depth of effort and results. So that's the first thing is make sure you have that named leader and that structure. The second thing, we talked a lot about it, but build the measurement infrastructure. It feels a little painful because you want to be doing all the fun stuff and just out there vibing. But the reality is you have to know what you're impacting at some point. And again, don't jump to measuring to make decisions right away, but do build the measurement system because you're going to need it. You need that telemetry, you're going to need to make sure that you can make decisions based on that information at some point in the future. And the third thing I might suggest is as a finance function, we obviously have a lot of great opportunity here to be sort of the tip of the spear and show what's possible when you automate a lot of these things and when you sort of build the future so we can be that internal proof point. So I would say eat your own cooking. If you come into an organization, say I'm going to show other folks the way by doing it here first. In finance, run the case studies yourself. You know, you look at something like an accounting org and at Zapier, it's one of our sort of marquee organizations when it comes to AI and automation because so much of that work is really transactional, repeatable this month end close process. And we've just, all of it is automated, almost all of it. And so it's a great way to say look, we've got accountants who are writing Python scripts. Like we, we can all do this. Let's, let's show the way with, with finance. So those are, those are the few things I would suggest coming into a new organization or trying to sort of inject this into your organization as a cfo. That's great. These are really great. Great. Depends. And as I was listening to you Ryan, I think something that strikes me when I'm listening to this overall episode and as we approaching like the end of it is it seems that you are playing a key role in the kind of like the culture of the company. You're not, you know, you know the type of CFO that is like on the side or just in your little office and just you know, letting other functions that drive the entire thing like how do you define and how can you give tips to other CFOs in the space to be like front and center as part of the culture of the company. And how do you drive that like in general? And the reason why I'm asking is that this is a trade that we've seen so many times on the growth minded cfo. But I know that this is not always the norm outside and you know, this is something that a lot of people want to do. So what would be your tips for new CFOs in the role around how you play your part in the culture of the company? Yeah, the first thing I'll say is join a company that has a strong culture. That's step one. And if it's not there, then you might be responsible for helping to build that. The lucky part for me is that Zapier is Built on an incredibly strong culture and reinforced on a daily basis. Our values are literally published everywhere, reinforced in slack across all these tools. And so I think that that by itself makes it easier. It makes it easier for me to sort of use that culture and lean into it and help push mandates as a result. But even still, to your point, many times CFOs they often are represented and act as the gatekeepers, they're the no, the know people. And I think especially in this new AI sort of era, this is also a very important part of the job. But I think there's a reframing here where we have to be the capital allocators, not just the gatekeepers. And that's part of that sort of culture mindset where as a capital allocator your job is to make sure that you are putting your resources towards the most impactful things. But as part of that, in this high paced sort of learning environment, you have to be flexible in that and show folks that it's okay to take risks and make mistakes and spend money where the ROI isn't entirely clear. The idea here is that it's up to us to build the thesis to say, look, we're going to spend X amount of revenue on these AI tools and we're going to learn and we're going to understand where it's useful and where it's not. We're not going to just sit here and approve and deny requests or deny them rather just because there's not a strong enough piece of evidence that suggests this is going to be useful in the future. So my advice is to just recognize the moment for what it is, act as that capital allocator, avoid the gatekeeper mindset and recognize that you as a CFO can influence folks perception of the culture by the way that you do sort of interact with those approvals, denials of this type of spend and these types of behaviors. That is a super, super, super interesting way of ending that episode, Orion, because we talked about culture but you linked basically like finance budget approval and stuff to a behavior that actually drives the right culture. And I think this is really inspiring for every emerging finance leader that wants to work in a company that is forward thinking and also kind of moving at the pace that Zapier is probably moving and also the pace of this episode because we had a super fast paced episode today with you Ryan, so many times my mouth, we have a value at zapier default to action and I'm always right at the top. We're here to get things done. We could definitely feel it. And this was maybe a question that I'll ask you offline around. How do you actually find out the right culture before you join the company? Because you mentioned that that's for another episode, but that's for another episode of the Growth Minded CFO Podcast. Ask lots of questions. That's my short answer. Love it. That's great. Thank you so much, Ryan, for being with us today. This was an amazing episode. I hope you liked it as much as we liked it. And if you liked that episode and if you want to hear from all other great CFOs like Ryan, please let us know. We're always happy to get your suggestion. You can also apply on the Growth Minded CFO podcast. We have more and more people applying now. This is really great to see, like the audience growing, but the audience is definitely growing because we have great guests like you, Ryan. So thank you so much for making the time today. You can help us as well by subscribing to the channels, liking it, and most importantly, sharing the episodes with colleagues or anyone else. There's probably great learnings that you want to share with other people. That's right. Did I say everything we needed to say before we close that episode? I think you said to subscribe, right? That's what we always forget to say at the end of every episode. Subscribe, subscribe. And we'll see you next time on another episode of the Growth Minded cfo. Thank you so much, Ryan, and thanks everyone. Have a good one.