Toast VP of Product on Shipping AI Into a 14-Year-Old Platform and Building for the Real Economy

Carlos González de Villaumbrosia 0:00 You guys had a killer quarter. Q2 was the fastest, steepest growth year in history, growing at 2.4 billion, 25% year over year, profitable.

Maggie Crowley 0:09 The better we are at understanding what our customers need, the better able we are to build things that they're willing to pay for.

Carlos González de Villaumbrosia 0:14 Maggie Crowley, VP of Product at Toast. Ultimately, I think you're building a product for a lot of people who are not on X, right? That's the real economy.

Maggie Crowley 0:22 Our job is to figure out how do we take all that technology that's coming out of Anthropic or OpenAI and put it to work for them so they don't have to spend the time thinking about, you know, which model do I use or how do I set up this workflow?

Carlos González de Villaumbrosia 0:34 This looks so similar to the kind of questions I used to ask my team every week, and now I just use Claude Gen and I'm a happier person.

Maggie Crowley 0:41 You can't necessarily launch something that doesn't work, then they're gonna not wanna use your products and you're gonna burn trust.

Carlos González de Villaumbrosia 0:46 One thing I noticed, especially with AI, is that the user is changing their expectation from, "Hey, tool, help me do this," to, "Hey, tool, do this for me."

Carlos González de Villaumbrosia 0:56 Hey, this is Carlos, CEO at Product School and your host on the Product Podcast. My guest today is Maggie Crowley, VP of Product at Toast. Two fun facts before we start. She finished fifth at the 2006 Winter Olympics as a speed skater, and she was contributing to Product School years ago back when she was at Drift. Toast is the point of sale behind the restaurants and coffee shops you already pay at, and her customers are not on X all day. They are running slim margin businesses that have zero patience for AI theater. Things we'll cover in this episode. Shipping AI into a 14-year-old platform,

Carlos González de Villaumbrosia 1:30 not a blank slate. 10 customers, one WhatsApp group, a year of iterating before GA. The least tech savvy operators loved it first, and why. How long you have before consumer patterns become your users' expectations. A live demo changing a real menu price from chat. Let's get into it. Welcome to the Product Podcast, Maggie.

Maggie Crowley 1:50 Thank you so much. Happy to be here.

Carlos González de Villaumbrosia 1:51 You know, I have two fun facts. One is related to your own career, and you told me that you were fifth at the Winter Olympics in, in, in 2006, right?

Maggie Crowley 2:01 That's correct.

Carlos González de Villaumbrosia 2:02 A pro skater?

Maggie Crowley 2:03 Yep, speed skating.

Carlos González de Villaumbrosia 2:04 Nice. And the other one, probably less impressive but still cool, is that you were a contributor to Product School many, many years ago, way before you were a VP of Product at Toast. I think at that time you were a senior director of product at Drift.

Maggie Crowley 2:16 Yes. Yes, I can't... It feels like not as long ago as it actually was, unfortunately.

Carlos González de Villaumbrosia 2:22 Few things have changed since then, and that's why I'm

Carlos González de Villaumbrosia 2:25 excited-

Maggie Crowley 2:25 Yeah

Carlos González de Villaumbrosia 2:25 ... to have you back. Uh, yes, so for some context for, for people, I'm, I'm sure everybody recently has used Toast at some point, but maybe they don't, don't even realize. So maybe you can explain what, what Toast is.

Maggie Crowley 2:38 Absolutely. So Toast is a retail and restaurant point of sale, and it's the thing where if you go to a restaurant or you go to a coffee shop and you wanna pay for your meal, you're probably swiping your card, whether it's at, you know, a kiosk or at, you know, a counter or maybe there's a handheld that they bring around where you swipe your card. Um, but it's... I think that's the thing that most people interact with, but behind the scenes, it's the hardware in the store itself and all of the software that helps our operators run their menus and their inventory and their te- their team schedule and, you know, their prep and how they understand how their business is doing. So there's a whole host of things happening kind of above store. Um, and then we also have a bunch of products that are for how we help our operators drive traffic, uh, to their locations, and then we have a consumer app as well.

Carlos González de Villaumbrosia 3:26 Well, that, those are the things I want to unpack. I mean, we spend a lot of time on X, and we are hopefully at the forefront on AI, of AI in some, in some ways. But ultimately, I think you're building a product for a lot of people who are not on X, right? That's the real economy. And so the way you probably build for them and drive adoption is different than some of the keynotes that we see from OpenAI or Anthropic.

Maggie Crowley 3:47 That's true. Yeah, but I, I think that

Maggie Crowley 3:50 I always, or at least I think about our role at Toast as these operators are super sophisticated. They run really difficult, slim margin businesses, and they do it largely because they love it. And our job is to figure out how do we take all that technology that's coming out of Anthropic or OpenAI or whoever and put it to work for them so that they don't have to spend the time thinking about, you know, which model do I use or how do I set up this workflow? But our job is to figure out how can we make it useful for them, and then how do we make it something that is actually real and is not a distraction? I think there's so much that you could do with AI, and there's so much theater around feeling like you've done something. But the question is, did it actually drive an outcome? Did it help you make more money? Did it help you save money? Did it save you real time? That's the bar that I think about when we're building with these tools for businesses that I think is just a different type of challenge that I find to be really interesting.

Carlos González de Villaumbrosia 4:46 So one, one thing I, I noticed you mentioned on, on X recently

Carlos González de Villaumbrosia 4:50 is that you are building agents within products that, SaaS products that existed before AI, right? So that's a key difference from maybe AI native products that didn't have something to compare with. So from your perspective, what does that really mean, and how is that different than an AI native product?

Maggie Crowley 5:07 I think part of the challenge is that you have-- There's a couple challenges. One is your customers, if you're lucky, or your users already know what they use you for. And so they have really specific ways that they have set up their workflows, whether that's just the muscle memory of how they use your tool or specific workflows in the tool that you have, and they're used to

Maggie Crowley 5:30 doing that. And so anything that you wanna do has to convince them to do that in a different way. So there's a, there's a level of, or layer of behavior change that you have to figure out. Um, there's also a sense of, you know, how much change can your customers absorb? Do they wanna try something new? Is the thing that your AI tool does actually germane to the outcome they're trying to get? And how do you make it useful for them? How do you make them wanna use it? Um, and then there's a scale question, which is you can't necessarily launch something that doesn't work, isn't fast, isn't accurate, isn't performant to your install base. Because, you know, then they're gonna

Maggie Crowley 6:05 not wanna use your products, and you're gonna burn trust. And I think if you're early, there's a little bit of a sense of you can just try things and test it out. I think we-- you, you talked about X. There's lots of things that we're seeing come through that channel that are early, that are a little bit broken, maybe the, like, integrations aren't great or the security isn't perfect, but that's okay because we're all early adopters. I think when you're building in a, in a platform like Toast, you have to think about the bar for quality excellence, how it works, is really high from the start. And so I think you have to get creative about how you build in that environment. It's not to say that you can't build quickly and iteratively, it's just a different way of thinking about it.

Carlos González de Villaumbrosia 6:39 Mm-hmm. So how are you building that environment?

Maggie Crowley 6:42 Yeah. So we-- The way we thought about it was in the very early days when we started working on this at, uh, the end of 2023, and when we started, we didn't even really know exactly what the use case would end up being, but we knew this technology was coming. We knew we wanted to start trying. And so what we did was we, we

Maggie Crowley 7:00 found just a handful of design partners, is what we called them, and we put ourselves in WhatsApp groups with these customers, and we just started turning an assistant on. We picked one platform, so we started on mobile, uh, just to make it easy for ourselves, and we picked kind of one use case around reporting, and we just shipped it to these 10 people. And we kind of walled ourselves off to try to iterate with them as fast as we could to get to a point where the product was good enough so then we could start to scale it. And we've, we've sort of done that with every major feature that we wanna release since then. Um, so we did that for honestly about a year, and then the models finally got good enough at the early 2025, and then we GA'd last fall in October, and then since then have built a lot into it. But it was really that process of finding a couple of customers, building really quickly and iteratively with them, being in a channel where you could get fast feedback, and then scaling from there is how we figured out kind of how to operate. And then I think probably the other thing that I would say that is unique about building in this space is that you have to bring the AI to your customer or your

Maggie Crowley 8:05 user. And I-- We couldn't assume that they would discover it on their own. And we-- You know, just adding a chat box somewhere isn't enough. Like, you have to bring it to them. You have to get them over the blank page problem. You have to help them understand when to use it. And an enduring challenge of these tools is that even if you get one of your users to use it, you're probably adding all sorts of capabilities in the background, and there's no way for them to see that, 'cause it's all behind the scenes. And so you have to constantly think about how do you, A, get them to use it, and then B, help them understand all the stuff that you're adding to it. And you have to think about constantly re-onboarding your users because it's constantly changing. And so it's just, like, a interesting way to think about a product and maybe before you could launch something and assume that once you, your user figured it out, used it once, they were-- they kind of understood how it worked. That's not the case with AI, and so you have to build a better

Maggie Crowley 9:00 muscle around bringing the product to them, helping them understand what it does, and then helping re-engage them when you have new things.

Carlos González de Villaumbrosia 9:05 I think that is a good p-point. I mean, onboarding has always been critical, but especially as now you start shipping more stuff, how do you-- Especially for someone who's probably doing other things, right? We go to bed thinking about that product we are building, but they probably have other things to figure out. So in your case, how do you do that type of onboarding differently now to make sure that they, that is driving the right level of adoption for those new, um, capabilities?

Maggie Crowley 9:30 Yeah. I think, I, I think if I had totally cracked the onboarding problem to an AI product, maybe I would, maybe I would have my own startup or something. I don't know. Um, I think the way we think about it is w- if the better, the more you understand your customer or your user and the way that they're working and what they're up to, the better you're gonna be at figuring out when they should use the thing you're building. And that, and that's not unique to AI. That's the same as any product that you'd be building. But I think you have to really understand

Maggie Crowley 10:00 physically where-- For us, physically, where are they in the store, in the restaurant? What time of day is it? Is it during shift? Is it not during shift? What role do they have in the restaurant or the retail shop, and where are they? And then what might they need to do? And then how do you close that gap and bring them whatever the, the use case is at that moment? Um, and so practically speaking, that's, you know, we have, of course, have this assistant, Toast IQ, uh, that's in our product, but it's, it's entry points throughout the product. Um, it's pre-written prompts that we know work really well, that we know are questions that they're gonna ask. Um, and it's making sure that it works really well on our mobile app as well as our desktop, and that we tailor the starter prompts or the mini prompts that we have set up based on what they're doing in those platforms at different times of day. And so just getting really specific about all of those things and trying to make it really easy for them to get started, 'cause I think once they get in and they start asking their questions, then it's relatively easy, but you just have to figure out how to get them in to start.

Carlos González de Villaumbrosia 10:58 And, and by the way, this is spoiler alert, but I know you are going to give us a demo on, on-

Maggie Crowley 11:02 Yeah

Carlos González de Villaumbrosia 11:02 ... Toast IQ. And, and, and I think that's powerful because ultimately proves, right, like how people are acting with your product. I, I-- One thing I noticed, especially with AI, is that the user is going, is changing their expectation from, "Hey, tool, help me do this," to, "Hey, tool, do this for me."

Maggie Crowley 11:18 Oh, yeah. I think, I think about this a lot because the dream is, of course, what you're saying, to actually hand off a workflow end to end. But there's-- So it's, it's interesting Muse came out, what is it, last week, and Grok bot maybe like a week or two before that. And so many of us are trying out all those tools and playing around with them, and it feels that those tools to me and those patterns feel inevitable. But there's some unknown amount of time from when the consumer market has those things to when they're fully adopted in a B2B context. It's, and it probably depends heavily on the type of user you have and the space that you're in. And so it's interesting to be in this moment where we see where things are going, we see where the behavior is going, and we have a chance to figure out how to do it well for our users. And I, I think to me it's the most interesting time to be building in a space like this, and I love building for a customer like ours because they have such a low tolerance for doing AI for AI's sake. And They're-- They don't care at all if it's AI or Toast IQ or whatever. All they care about is, "Did you help me with my business? And did you help me with a specific problem?" And so I'm just really interested to see how it all plays out with these patterns and how they're gonna show up in that place.

Carlos González de Villaumbrosia 12:24 By the way, you mentioned Muse and Grok Bot. I've been obsessing over Instinct. Literally-

Maggie Crowley 12:29 Okay

Carlos González de Villaumbrosia 12:29 ... yesterday I went to sleep at 3:00 a.m. just because I couldn't stop it. You know? Like, the thing kept doing things for me, and that was, um, fun and addictive, and it also made me miserable. But

Carlos González de Villaumbrosia 12:40 I can see that B2C pattern now that is making someone invincible, right? Like, can eventually apply-

Maggie Crowley 12:47 Mm-hmm

Carlos González de Villaumbrosia 12:47 ... to B2B and allow business owners of all kinds to have a much better understanding of their business and eventually get to do things that they used to do when the business was smaller.

Maggie Crowley 12:57 Absolutely. And I think there's almost a clock because the second you start to see interactions like that in the consumer market, there's a certain amount of time until those patterns become standard. And then if you have an AI product, your users expect that. And so I also see it as, okay, we're seeing these patterns happen now. How long do I have before my users assume that that's how my tool w- will work? And so I have that amount of time to make sure I can do whatever it is those things do, and we have to stay rele-- stay up enough to be relevant in the space. So that's just another interesting angle around building with AI right now that I think about. And it, it's fun because to me it means that you almost have no idea where your product is gonna go, and you have to figure out how to be more flexible because you don't really know what the patterns are gonna be, A, in terms of the interaction patterns, but also B, in terms of what the models can do.

Carlos González de Villaumbrosia 13:46 Yeah.

Maggie Crowley 13:46 And so before, I think before AI, building was almost getting boring, relatively, and now it's much more dynamic and interesting because we have no idea what's going on.

Carlos González de Villaumbrosia 13:54 Totally. You, you don't know where the limits are, right? Before, like, the buttons would define your behavior. Now you're like-

Maggie Crowley 13:59 Yeah

Carlos González de Villaumbrosia 13:59 ... "Hey, can you create this new button and do something else?" [laughs]

Maggie Crowley 14:03 Yeah.

Carlos González de Villaumbrosia 14:03 But-

Maggie Crowley 14:04 Totally

Carlos González de Villaumbrosia 14:04 ... I, I, I also want to cover the market landscape because as someone thinks about their restaurant, of course, point of sale is important, and I imagine companies like Square, Owner, are part of that, um, market. But even broader, as I think about food delivery with companies like DoorDash or Uber Eats or table reservations with OpenTable or reputation management with Yelp, like, how are you thinking about your new position in this bigger platform?

Maggie Crowley 14:30 It's a good question. And I think everybody is trying to figure out how the market is gonna play out. I don't know that anybody knows the way it-- we're all gonna interact and the way it's all gonna land. I think, at least, like, the way I think about this problem is, how do we make sure that we're keeping our customers at the heart of what we're doing? And I think it's so easy to get distracted by that and to get distracted by the competitive dynamics, the market dynamics, the who's gonna be the winner? Is there gonna be a winner? How is it all gonna play out? And it's interesting to game that out from a strategic standpoint, but at the end of the day, the person who wins or the company who wins, if there is such a thing, is the company that does the, solves the right problem in the best way for their market and deeply understands that. And so I think now even, even more than before, the most important thing to do is to really understand the user and what they're up to. And I mean, like, be with them physically. Go to their places. Watch them do their work. Embed with them. Be a person that they can call. And the more that you do

Maggie Crowley 15:40 that, the better you're, positioned you're gonna be for whatever that future's gonna hold. Because in the same way that three weeks ago we had never heard of Muse and then now it's all that we can talk about, or same thing with Instinct if that's the one that you're using. Who knows what that's gonna be, and who knows what new entrants are gonna happen or what, what other companies are gonna do. And I think if you try to anchor too much on gaming that out, you're gonna miss the point of n- right now, which is build for your user and do everything you can for them because that's the thing that's gonna make you successful no matter kind of what happens in the market.

Carlos González de Villaumbrosia 16:08 I'm looking at some of the market numbers.

Carlos González de Villaumbrosia 16:10 You guys had a killer quarter, right? Q2 was the fastest, steepest growth curve in history. Uh, growing at 2.4 billion, 25% year over year, profitable. I mean, at the same time you also had to be contained, right? So from a strategic standpoint, how are you figuring out your, your growth levers?

Maggie Crowley 16:30 Again, it's like the same-- It's as simple as what I'm saying. I think it's the better we are at understanding what our customers need, the better able we are to build things that they're willing to pay for. And especially in a business that is as tight as a restaurant, you don't have the option of just assuming people are gonna pay for extra things. You have to understand the P&L of a restaurant to understand where is any additional dollar that you're gonna charge them gonna come from, and is that actually gonna be worth their time? And especially being a public company, we have a really strong finance team who's amazing, and we've been really disciplined about that. And so the way that we think about building, it

Maggie Crowley 17:05 always has to be something that really matters for the user. And so that's the way we think about growth is, how do we deeply understand what they need and how we solve real problems? And then the rest kind of figures itself out in the way that we approach building.

Carlos González de Villaumbrosia 17:16 One of the, the things I, I read about kind of the positioning statement of your new product, um, Toast IQ, is that you're going from we manage your point of sales to we grow your restaurant, right?

Maggie Crowley 17:27 Mm-hmm.

Carlos González de Villaumbrosia 17:27 So you are now taking a bigger stance at helping your client make more money. So-

Maggie Crowley 17:33 Yeah

Carlos González de Villaumbrosia 17:33 ... I'm curious about then what does that really mean for other tools, right? Email marketing tools or other technology that maybe is not restaurant specific, but in a way are now being more related to,

Carlos González de Villaumbrosia 17:45 to what you do.

Maggie Crowley 17:46 Yeah, the marketing one is really interesting. That product has been so successful. And I mean, what's, what's interesting about it is that there's, I think there's one customer that has reduced their marketing agency spend by over 70% and can tie, um, marketing to sales and attribute a specific amount of sales to the marketing site since they've launched it via this product that we built And so we have thing-- examples like that that show that you can use AI to dramatically reduce the cost of doing something, but also take off work that you're not willing to do yourself or you don't want to do yourself. 'Cause a lot of we-- A lot of the, the sweet spot for us is, you know, a marketer or a restaurateur isn't necessarily someone who loves doing email marketing or online marketing or building a website or whatever. And so then they pay someone to do it, and then that's really expensive, and it's a low-margin business, and so they're happy to pay less than they would pay an agency to get a lot of what they're getting out of the agency, but using AI. And so that's, that's been one of the breakthrough use cases for us for, for some of this technology. Um, and that team has done a really amazing job of building an agent that really solves a painful problem of driving demand for a restaurant.

Carlos González de Villaumbrosia 18:57 So this is an agent that can do a lot of the marketing stuff on behalf of the restaurant via email, text, and other channels.

Maggie Crowley 19:03 Yeah, it's called Toast IQ Grow, and it's just the whole game of that is how do we, you know,

Maggie Crowley 19:10 put butts in seats and all the different pieces that go into that and helping our customers offload that work.

Carlos González de Villaumbrosia 19:16 And the, the other big thing, as you mentioned marketing, is, uh, AEO, right? How you show up not only on Google Search but also now in LLMs.

Maggie Crowley 19:26 Mm-hmm.

Carlos González de Villaumbrosia 19:26 So that's, again, super sophisticated, requires ton of effort, and someone who's running a restaurant maybe is not focusing that much on, on that online positioning. So how are you thinking about that as it applies to how you're building product?

Maggie Crowley 19:38 I think it-- I mean,

Maggie Crowley 19:40 I feel like I'm a broken record here. It's the same thing. It's just the, the more that we can spend our time really understanding the technology, the better we are at doing that for customers. And so the way we structure our teams is we try to make sure that we have a PM focused deeply in that kind of space, so that then they can figure out with their technical counterparts and bring that solution to the, to the users that we have, rather than forcing a customer to try to figure it out themselves. And we're doing that across basically every use case we can think of, um, and trying to figure out, you know, what can we take off your plate with AI? What can we do? What problems can we solve, and how do we make it easy for you so that as a customer, you don't have to think about, you know, what model do I use? What tool do I use? Should I use Anthropic or should I use OpenAI or should I use an open-weight model? We can do all that for them, so all they have to do is just plug in Toast and it works.

Carlos González de Villaumbrosia 20:31 For you as a VP of Product, you are not an IC anymore, but I imagine you are also carving out time to do some IC work, stay in touch with customers. Like, how does that day-to-day look like now different than, you know, before AI became that popular?

Maggie Crowley 20:46 I really think that in this day and age, if you don't

Maggie Crowley 20:50 keep a hand in, in how the teams are building, it's gonna be really difficult to be a leader in nine months, a year or two years, because the way that the teams build is so different. And so whether that's finding a project to do on your own or finding a way to be a line manager versus being a manager of managers or whatever level you're at, I think that's super critical for leaders right now. For me, the way I've done it is any time a, a PM, you know, maybe goes on parental leave or we have an opening, I try to hop in and do that job for a little bit to keep, keep myself in the weeds on what those products are, and then also make sure I understand what's going on on the sort of frontline level. And then I have areas, Toast IQ as an example, where I don't have a, a middle management layer and I manage the frontline team myself because I wanna stay close to the product. And I think you can scale yourself as a leader much more effectively using AI tools to allow you to do that and allow you to keep a flatter team than you used to. And I think to me, if you don't do that as a lead, there's only so long until you, you just don't understand how these tools work, and you're gonna be probably somewhat irrelevant.

Carlos González de Villaumbrosia 21:54 Well, case in point, I would love for you to roll up your sleeves and, and show us your product.

Maggie Crowley 21:59 Okay. All right. So I think I have to imagine there's the series of caveats that we all share about this is a demo environment, and this is AI, so let's go on a journey. Um, I, I think the, the most useful thing to think about as we go through this is that I'm gonna show kind of one typical type of, of workflow that our customers go through, and there's of course tons of other things that we can do. Um, but I find that most customers usually like to start with just trying to understand what's going on, um, especially like what's going on today, what happened yesterday, you know, where am I at on my week, um, how do I maybe then how am I gonna react to what's coming next. Um, so I might wanna like understand how that compares to the week prior. And you and I were joking about this maybe before we started recording. This is where we work on our comedy routines as we wait for AI to load, uh, day in and day out-

Carlos González de Villaumbrosia 22:56 Oh, yeah

Maggie Crowley 22:56 ... when you're demoing these for your customers.

Carlos González de Villaumbrosia 22:58 [laughs]

Maggie Crowley 23:04 All right. So then it

Maggie Crowley 23:05 can tell you, you know, how does this compare to the week prior? Maybe there's something specific that you're seeing. Looks like for this, in this instance, my average check is higher. So let's see. Um-

Carlos González de Villaumbrosia 23:17 It's crazy

Maggie Crowley 23:18 ... maybe what drove that.

Carlos González de Villaumbrosia 23:19 This, this looks so similar to the kind of questions I used to ask my team, uh, every week, right? Like, we'd go, and then someone would have to pull the report from Salesforce, and there was a meeting for the meeting, and it was just so horrible.

Maggie Crowley 23:32 Mm-hmm.

Carlos González de Villaumbrosia 23:32 And now I just use Claude, and it's connected to my Salesforce via MCP, and I ask those questions, and I ask a follow-up question, and I'm a happier person.

Maggie Crowley 23:42 Totally. And I think a lot of these-- Some of the, the really interesting magic when we first started is that it, it wasn't necessarily that you couldn't do this or you couldn't answer these questions if you wanted to in Toast, but maybe it just wasn't the easiest to do it. Maybe the way, it wasn't intuitive for them to find it. Toast has been around for 14 years. We have so much product, um, and you know, a customer m-might not wanna spend the time learning all of that. And so this is just such a fast path to getting the answer. And what was really interesting is that I also learned in the very early days that we had this assumption that, um, you- Our customers, only really tech-savvy early adopters would wanna use this. But what happened was we found that some of the less tech-savvy operators loved it 'cause they didn't have to learn, learn software, and they just used their human language, and they loved it 'cause they could just ask questions without needing to feel like they needed to learn a tool. Um, so I always thought that was an interesting learning from the early days.

Carlos González de Villaumbrosia 24:39 Okay, so what do we have here?

Maggie Crowley 24:42 So it looks like the drivers I have are this prepaid dinner and my pepperoni flatbread. Um, so then I'm trying to remember common, like, what flat maybe, like, can you list out all the flatbreads on the menu? 'Cause I'm thinking about maybe I could juice my sales if I add to this menu category. And so a lot of what our customers are doing is trying to establish kind of day in and day out how their menus might wanna move, especially if there are changes in the cost of ingredients. And so one of the things that we hear all the time is that the cost of the ingredients and cost of food is going up, and it's really difficult to continue to raise your menu prices to match. And so they're thinking about what are other ways I can construct a menu that's gonna work, that's gonna help my business be successful, and take into account the fact that food prices or food costs are rising. Um, so it looks like we've got eight flatbreads. Maybe then I'll say, um, can you suggest-

Carlos González de Villaumbrosia 25:46 So Maggie, is the-

Maggie Crowley 25:46 What's that?

Carlos González de Villaumbrosia 25:47 ... plan that users use that interface, or are you also thinking about creating an MCP so they can also access the same kind of information from Claude or Codex?

Maggie Crowley 25:58 That is the million-dollar question.

Carlos González de Villaumbrosia 26:00 [laughs]

Maggie Crowley 26:00 Um, right now we're focused on how do we help our customers just get their work done in Toast. I think there are all kinds of examples of every SaaS platform asking the same question and testing out different things. Uh, I don't know that we have a perfect answer to that yet. Um, but again, I, I'm listening to our customers. I'm having calls about how, what information they wanna bring in, what information they wanna bring out, um, and how we help them get their work done. So-

Carlos González de Villaumbrosia 26:26 Right

Maggie Crowley 26:26 ... I don't know. [laughs] That's probably as far as I can go on that before legal comes for me.

Carlos González de Villaumbrosia 26:29 [laughs] As I see you do this digging, one of the things that come to mind is that the same way you, you are also integrating with a lot of the front-end tools like food delivery or, or, or, or restaurant reservation, you also integrate with a lot of the backend stuff, right? Because there's a lot of financial information here.

Maggie Crowley 26:47 Yeah. So we, we have partners that we work with across basically the full life cycle of a restaurant or retail shop, and we have products within Toast that they can use. Um, and then we also try to make sure we can pull in any other context that they have. Um, but again, our goal is just how do we help them get the information they need to get the job done? And in- increasingly with Toast IQ, we're trying to take that job as far as we can and take on as much of the work as possible. And so we end up in, in situations like this where I think in the early days, maybe just having insights was enough, but of course that's not enough anymore. You have to actually be able to make, make changes for a specific customer. Um, so maybe let's make it $14.99. That sounds good. And so this is something that before AI would've been relatively annoying maybe to do, and is, is the kind of activity that maybe if you're just making one menu change is, is relatively simple, but if you might, might wanna make bulk changes and make a lot of changes or maybe you have multiple menus across multiple locations, you can see how this kind of experience would get really complicated. And Toast IQ makes it a lot easier to just get it all done in chat. And one of the interesting things that's happening, um, is that this, just to give a little behind the scenes, this card is generated on the fly by Toast IQ based

Maggie Crowley 28:05 on the activity that you're trying to get done. And so we're using different frameworks to make that possible, and it's really interesting to kind of... It's been interesting to see how this comes to life because we built all these components in the backend, and then, or in the system, and then Toast IQ assembles it based on the question that you're asking, the thing you're trying to get done. And so it's cool to see this, this actually happen live because it's pulling together something on the fly and isn't something we've constructed to look ex- explicitly like this.

Carlos González de Villaumbrosia 28:31 And, and as you said, it's not just about getting the insight. You're also making actions that are being pushed now in real time to menus in this case. But I can see how you can also run a campaign, right?

Maggie Crowley 28:43 Totally.

Carlos González de Villaumbrosia 28:44 Or investigate what's going on with a customer or with a specific location. Kind of reminds me of e-commerce stores, right? Like e-commerce, like you are running this in an interface, and you're trying to understand what's happening with your inventory, and you are... And, and now you can apply that same mindset now to a, to a restaurant or a chain of restaurants.

Maggie Crowley 29:01 Exactly. So yeah, you can imagine doing pretty much any type of activity in this way, um, whether that's, like I mentioned, a bulk action, one action at a time. Um, you can schedule these kinds of things based on an event, um, or a specific timeframe. And so there's lots of different things you can do, um, to take work off our customers' plates with this tool.

Carlos González de Villaumbrosia 29:21 Well, Maggie, thank you for letting me put you on the spot here and, and demo for real, uh, something that is, that is cool and that you, and you've built. So I- I'm excited. I, I like the-

Maggie Crowley 29:30 I mean, also, like, m- major credits to the team. Yes, I've been involved in this since, since the beginning, and I was the first PM or this maybe the second PM that worked on it. But... And there's, now there's a big team that works on it, but yes, they've done an incredible job on this.

Carlos González de Villaumbrosia 29:42 But hey, here you are, you know, sharing a screen and going for it. And, uh-

Maggie Crowley 29:45 [laughs]

Carlos González de Villaumbrosia 29:46 ... you know, I typically host product leaders from OpenAI, Anthropic, and there's nothing wrong with that, but I think it's something special to also bring some of this magic to the real economy.

Maggie Crowley 29:56 Yeah, and I think it's, it's-- we talked about this a little bit, but the... It's so fun to be able to take that and bring it to a group of people who maybe wouldn't typically be using that kind of software or maybe aren't gonna be the most sophisticated in using it or, you know, aren't the early, early adopters. Because I think the bar for getting it right is so much higher, and you, you can't, you can't just assume they're gonna use it 'cause they like trying cool new stuff. Um, so it, to me that makes it as a more interesting challenge from a product perspective.

Carlos González de Villaumbrosia 30:24 And as you said, these businesses are usually like, have razor-thin margins, right? So, like, the impact you can make by optimizing certain things is, is real. It's, it's money in their pocket.

Maggie Crowley 30:35 Yep. Absolutely.

Carlos González de Villaumbrosia 30:36 Cool. Well, thank you for your time.

Maggie Crowley 30:37 Yeah, of course. It was fun to come on and chat about Toast IQ. [outro jingle]