The Procter & Gamble Company ($PG)

Great. Good morning. I'm Robert Ottenstein. I head Evercore ISI's Global Beverage and Household Products Research. We're super excited to be starting day 2 with Procter & Gamble. I think everybody is well aware that Procter has over 20 billion-dollar brands. In fact, I think they've stopped counting how many at this point. But people may not be quite as aware of the fact is they also have supporting that in the back, hands down, what is generally acknowledged as the most reliable supply chain. Moreover, and enabling the supply chain and increasingly enabling the brands at the front end and the back end is also an IT capability that is widely regarded as the envy of the industry by anybody that you talk to. In that regard, we're super excited to have Seth Cohen here with us. Seth runs the Information Technology Office, Chief Information Technology Officer at Procter. He joined Procter in April of 2024, having come from a similar position at PepsiCo and at Reckitt. So I don't think there's anybody on the planet who has run 3 world-class information system programs. In his capacity, he leads the digital transformation. He oversees technology strategy, data, AI, cybersecurity and really the digital capabilities around the world. Joining him is Keri Cowan. Keri ran China Hair for Procter & Gamble, I think, for the last 3 years. She's moved over now as Senior VP in the IR department and is taking the helm there, if not already, very shortly. I'm not sure on the -- exactly on the timing on that. So super happy to get started here. So Seth, you're still fairly new to Procter. As you came in the door and you started to kind of get a sense of the levers and the muscles of Procter on its digital capabilities, what was your early assessment of the capabilities? And what are your focus areas over the next 2 years to help deliver the strategic vision of the company? And just to kind of hone in on it, if you could talk about maybe 2 or 3 key goals as CIO.

That's a great question. First of all, thank you for having us today. I'm excited to talk about journey, where we are and the great work that's been accomplished and what's left to be done. As you commented, I've had the opportunity to work at other big, big brand companies. And coming into P&G, one of the things that I was quite impressed with, and this isn't maybe the limelight of AI, but I think it's the most important part of AI, is P&G's data capabilities. P&G is known in the industry of having some of the most standardized systems of records of any of our peers. So when you think about like SAP, again, not terribly exciting, I guess if you're an SAP person, you think it would be, but in general speak, not terribly exciting, but yet we have a single instance of SAP globally that runs all of P&G. And while from an application lens, there are some advantages, from my view; from a data lens, it's a superpower. Because if you think about the data that then is structured underneath that, the transactional data, it is 100% common. And therefore, when we start to then feed the AI, and I joke around, AI without data is simply A, it's artificial, it really starts to drive new capabilities. So we've been focusing on harnessing that data into a core data lake, as well as the other piece that we're now able to do because of that foundational capability is our AI factory, which is where all the models hit all of the digital products can be built upon sit there, we can scale. In terms of where we want to go and where we're focusing, I would say 4 key, I'll call them, toolboxes. Customer and consumer is first. And there, we're focusing on integrated brand building. So you think about all of the concept to creative to media as well as enablement of the sales force, is a big, big piece there. Second is supply chain. Supply Chain 3.0 is a big enabler for us from both a productivity [ especially ] as well as a reliability angle. R&D is #3, and they're focusing on how AI can make us faster in both creating insights as well as generating new products. And lastly, internal efficiencies. And that's a big, big area for us that we're driving toward.

Just for those who are new to some of the lingo, can you just briefly explain what exactly is the data lake? It sounds kind of fun and rural and nice. What does it actually mean? Why is it important?

It's not necessarily a place you go for your suntan. But what a data lake does for us is, if you think about what can happen when you have individual systems that are out there, the data can become quite siloed. So if you want information related to production, you might go to an SAP system. If you want data related to a sales force, you might go to Salesforce, et cetera. The problem with the slowness of the data is if we're looking at creating digital products, we want to string the data together to let the AI start to go across functional boundaries to try to solve business problems more from an end-to-end point of view. A data lake is the technical capability of allowing us to bring the information ultimately together.

Got it. Great. Great. So one of the things that investors are very much focused on, right, is we tend to live in the very short term. We get the scanner data that comes out every week, every 2 weeks. And although we're supposed to be looking really long term, people look short term. And what seems to be happening in a lot of industries, including yours and particularly in areas like beauty, is that the smaller so-called digital natives who don't have any of your R&D, any of your supply chain, any of your muscles, but they do have some savvy, right, in terms of dealing in the digital landscape, in aggregate, they seem to be winning in many ways. So I was wondering if you could talk about kind of the dynamics of competition on Amazon with the digital natives and how you're using these capabilities to meet these new sorts of challenges. And maybe talk about the difference between competing on Amazon versus Walmart.

That's a great bunch of questions. So let me try to take them one at a time. So I want to lead with we are absolutely also looking to get inspiration to learn from where others are succeeding. We're dealing with spaces that are evolving very rapidly. So please don't treat anything that I say as we got this solved pencils down and we're moving on. So we are looking externally all the time to see who's winning, who's doing things that maybe we ought to think about doing it in a more scaled way. When we talk about beauty specifically, and it's a fascinating category, coming from my last few companies, I would say beauty was probably not as focused as it is here. It is a very unique category in how the consumer ultimately engages into the category. And as you rightfully call out, a lot of small players are in that space. Now the thing that just to be balanced with is I think Nielsen is a number that -- please don't quote me, it's somewhere around 95% failure rate for the small players in these spaces. So as we look at these different providers in these spaces, we don't want to necessarily emulate a 95% failure. But there are learnings that we do want to make sure that we are both embracing and scaling. And I would argue, scaling is the name of the game. If we -- if you're doing just a series of pilots, it's not going to be materially impactful to the company. So from that, I would say some of the inspirations that we've gotten, and we've already built in now into the beauty categories and now we're scaling to other categories, is this whole consumer journey, meeting the consumer in the generative AI spaces that she's playing in, meeting the consumer in social, leveraging different vehicles for that consumer to engage with us, whether it be channels that we are authoring or key opinion leaders, they call them KOLs, which we are using to help enforce the capabilities of the product. And then, of course, user-generated content is also a big, big play. So we have a lot of activities going on in scaling the abilities to do that. Your question around Amazon and others, the thing with the retailers that is great, in my opinion, is that we actually all have a very common objective. And that is we want to meet the needs of the consumer at the end of the day. We have a saying at P&G, and I love it, "Consumer is boss." And it really is the DNA of the company. It's everything that we ultimately do is about the consumer. We define what we call 5 vectors of superiority that we feel if we can meet 5 vectors of superiority with the consumer, we ultimately win with the consumer in beating his or her needs. And those 5 vectors are around a superior product, in a superior package, with superior communications like media, as well as superior sailing, so the product is available where it's supposed to be, and at a superior value. So we feel similarly to what Amazon would say or what Walmart or whoever we deal with and say, if we can meet those 5 vectors of superiority for the consumers that are shopping in those channels, we ultimately have met the needs that we're trying to do. And we're constantly evolving these and trying to improve upon these. So back to your question around Amazon, how do you win on Amazon? It's no different than if we can meet the 5 -- we call 5 vectors of superiority for the consumers that are shopping on Amazon, we feel that we're in a good place to win. And that's what we focus brand by brand with the different retailers, on making sure we are, in fact, meeting these. Or if we're not meeting the needs, making sure we have the right interventions in place to meet those needs.

Great. So look, if you walk into a Walmart, you're all over the place, right? You can't avoid Procter & Gamble. You go on Amazon, you need to be on the front page, right? That's super important, right? So one of the things that we understand or have been told at least is that Amazon has changed the algorithm a little bit and maybe constantly does so in terms of how do you get on the front page, how do you get that visibility. Can you talk a little bit about that? And it's probably a lot of confidential state secrets, but how do you try to game the system, or even in -- not even game it, but just make sure that you have a fair representation and that other people don't kind of do in rounds around you to kind of have outsized presence on Amazon?

No, it's a great question. I'll focus more on what we call the organic side of the equation, meaning you're not looking at a paid advertisement from us, but rather you've done a query on Amazon. And I would say this is probably similar to Walmart or any of the dot-coms that we deal with. The algorithms, I can't really speak of, right? I'm not privy. Would I like to know? Sure. But it's not necessarily something that they're open to share with us. But back to this idea of 5 vectors of superiority, Amazon or Walmart or whoever we're talking, Tesco, you name it, they want to make sure that they're meeting the needs of their consumers. And so what they're leveraging is their data sets to figure that out. And what I mean by their data sets, things like ratings and reviews, things like the product description page, so the information we provide, are highly valuable in correlating -- and sales, of course, in correlating the consumer question and search to ultimately what gets presented. So what we do focus on with Amazon, Walmart and others, is because we have a very large proprietary consumer behavioral database, we're able to now inject our insights and learnings from the consumer, what they're looking for, into our product descriptions. So ultimately, the consumer good companies are feeding the product descriptions of our products in, so that when the consumer asks what is the best, I don't know, razor to use, we're able to take that question and ensure that in a very easy-to-understand language in the product description page, that's included. So that when their algorithms are searching for answers, we're able to come up with the right answer. Of course, ratings and reviews are huge for everybody. So we want to make sure, if there are great stories to be told, we want to make sure we're telling them. We also want to make sure if there's maybe not so great stories to be told, we're reacting and understanding how to adjust so we can get back to that 5 vectors of superiority. So I don't have an answer from the algorithm necessarily. But I can tell you, we do a lot of matching of what we call generative engine optimization, which is, in the old days, we used to see SEO search, now it's more generative search, to make sure we have a strong match for what they're asking for to ultimately what data they can actually pull out of the details to provide to the customer.

Great. Great. So one of the questions that has started coming into myself and Javier, who covers Procter with me, is what is agentic marketing? What is it? Is it good, is it bad for Procter? Is it good for Walmart? How does it change the game? How do you use AI to deal with it? And I think a lot of the questions at the core, I don't think people really understand what it is, and it's, I guess, it's developing, become a big buzzword. So maybe you can enlighten us a little bit in terms of what is agentic marketing. How does this change the consumer path to purchase? And what are the new challenges and opportunities that it brings?

It's a great question. And if you have a definition, I would love to hear your definition as well. I think agentic is a -- it's an interesting term. And I suspect if I were to go around the room and ask all of you to define agentic, I might get slightly different answers by individual. What I'm going to focus on now is more the agentic kind of path to purchase. I think -- I'm hoping later we'll talk about media at some point. But the realities are, thus far, we're not seeing this idea of I'm going to let my agent just buy for me. And while this is not necessarily settled yet, we're not sure how far this will go, an example I would give from past days that gives me reason to believe that the human will still be involved is if you think about the subscriptions that we are often asked to subscribe to products, how often in the past have we subscribed? When you think about it, subscriptions are not that different from agentic in terms of it's an automated workflow that does suddenly pumps out products to you on a regular basis. But most humans are not comfortable even in that very specific use case to do it. I'm still of the belief that we're not sure how far this agentic workflow will take on. I do believe that agentic, it already is and will continue to grow as part of the workflow. And so what we focus on for that point is we want to make sure that from a consumer journey perspective, and more and more of the consumer journeys are starting usually either in the wild, when I say in the wild, like ChatGPT or Gemini, or within the walled garden of a specific retailer such as, well, Amazon used to call it Rufus, they just renamed it to Alexa or Sparky at Walmart, we want to make sure that when questions are asked, that we're able to understand that question and make sure that our information is being presented in a very accurate and thoughtful way. And so we have a lot of activities at scale that we're deploying around all of our categories where we spend time on what we call GEO search, which is this concept of generative engine optimization, understanding where are the engines going to seek out information to then make sure that we are presenting the right information for it to come back. And then from that point, to ensure, back to the earlier discussion, around this idea of making sure though that our product information catalog or product descriptions in the retailers, in whatever sites that we're selling in, have a clear match to it. So these engines, these LLMs have an easy way to match. And so that's where we're really focusing a predominant amount of our effort on, to ensure that there is that cleanness in that, I'll call it, accuracy of how that question turns into an insight for the consumer then turns into, hopefully, a purchase of our product.

Great. So if things weren't challenging enough already, at the same time this is all happening, right, media is proliferating like crazy, and the lines are blurring between what's a retailer, media, you've got influencers. I mean it's crazy just how complicated things have come in the last 5 years and kind of moving at a very fast rate. So that media proliferation and how that changes the consumer path to purchase is something that Shailesh has called out at conferences. So one big challenge, I'm sure it's a big part of your mandate and working on the marketing side, how is Procter responding to that environment? And how can your capabilities help Procter deal with this rapidly changing media world where lines are really blurring between retailers and media and everything in between?

You're right, it's a very fast-moving space. And from a consumer lens, and I'm sure -- we're all consumers. That's the beauty of working in this industry is that we can all relate to the journeys that we're talking about, is it's a lot of information is often being thrown at us, whether it be in doing searching or whether I'm on TikTok or I'm on Facebook or wherever I'm playing, to the number and potential touch points that could be there are exponentially different than the past. So what we're focusing on in this space, and I'll kind of take a walk down memory lane, before the explosion of social, consumer good companies might be able to get away with maybe 1 to 4 updates on the ads on linear TV for the year and be absolutely fine. Now we're dealing with needing to deal with anywhere up to 10 to 200x that number to be able to engage with the consumer wherever she may be walking or wherever he might be looking for products that are out there. So there's a couple of elements to this. One, I've mentioned is this whole degenerative engine optimization element. And that is quite important for us to make sure that we're staying on top of, to make sure that when you ask a question about a product, especially if it's a product in our categories, that we're able to give a thoughtful response through the engines that are out there. And I think that's a big, big unlock. We now have also layered in, and we're scaling this across every one of our categories, this idea of, well, then how do I generate 10 to 200x that content depending on the category needs? In the old days, you would leverage agencies. Well, the reality is, and we've talked about this, in some categories, the volume that we need to get to and the scale we need to get to, it's not realistic to assume an agency would be able to meet those needs. So we're internalizing some of the agency capabilities, specifically around media concept to creation, leveraging generative AI. And then once I get to creation, adaptation of it. And this -- it might not be well understood, but just having the asset is good, but the problem is, is that every site you go to has very specific requirements of that asset on their site: the size of it, the color pallets, et cetera. So baking that all in into an automated workflow is critically important for us. Taking that into the next level of, okay, well, now that I have this asset that's been sized for a specific location, what do I do with it? Well, we have tools, and we've in-housed this over the last few years, where we have media buying tools that are, I would argue, best-in-class. In fact, compared to where we were when we were using external help for this, we are seeing a tremendous higher impact at a lower cost for us to be able to do things. So the same ad is able to be presented and targeted to the right consumer base at the right time of day, at the right, I'll call it, purchase inflection point, to be able to do things. And then the round trip of it is measurement of performance. And you mentioned earlier, that you're 100% right, the lines are blurred. Used to be very clean. I have media companies and I have retailers. Well, now retailers are becoming media companies. And arguably, social is becoming retailers. Think about like TikTok Shop as an example. So to be able to see the attribution of that ad that was seen all the way through to a purchase decision so that we can react, in the old days, it would take us 4 or 5 days to see that, we now see it in, I'll call it, near real time, not 100% real time, so that we can then quickly react to that to adjust that workflow as we move forward.

Great. Great. So one of the things that has really proliferated is influencers. And there's -- and it's proliferated so much. There's macro influencers, there's micro influencers. I mean who knows how that's being segmented, right? And we have seen in some cases where that hasn't worked out so well for some companies in the beer industry, which I don't need to mention. But look, how do you deal with these influencers? And I don't even know how many you have. I mean, I think some companies we talk to, it's in the multiple thousands, like 50,000 in some cases. So maybe if you could talk a little bit about how you help the marketing to manage influencers, impact on brand equity. And to the extent that it's possible and relevant, maybe contrast how the influencer ecosystem in the U.S. may contrast with what's in China. Because it's been very big in China as well, where they call it KOLs, whatever. And it's a different -- it may be a different type of thing. But would love to get your thoughts on that.

It's a really good question. So KOLs, key opinion leader, is the term that we're using internally for this capability. And let me first try to paint out the different levels that you would have in these spaces. So at the, I'll call it, the highest level or the most controlled level, we have the content that we're putting out ourselves. The next level below that would be what we call these key opinion leaders. And these would be the few but very influential people, have lots of followerships and lots of [ influence], that we would contract with to get them to enforce brand messaging for us on behalf of the folks that they represent. When you mentioned that was in the thousands or tens of thousands, that's when we start to get into user-generated content. And that is also part of the equation for us. And there's different mechanisms to get user-generated content. One is just pure organic, someone just absolutely loves Old Spice deodorant and wants to scream from the mountaintops how much they love it. Hopefully, if you guys like it, you'll do that for us. But others might be us nudging it. So for example, we have loyalty programs that we will occasionally put out messages, hey, if you like this new product and are willing to talk about it, tag us and you might be put in a lottery to win something, et cetera. So we have these different archetypes that we're looking at in terms of who we get to enlist to talk about our brands. The most important point, this is the piece that we've now really ramped up, this is one of the key learnings that we had. You asked earlier about, hey, what happens in beauty when some companies are doing things. One of the early learnings that we had was we were not aggressive enough in the measurement space in this thing. We've deployed this now and we're actually seeing some great success. But we're now able to see in near real time how the KOL performance is. And we're looking for a few things. First, are they on message? So we're using generative AI to tag and understand if they're on message or not. Second, are they getting a level of followership that's giving us a signal that this thing could become a viral communication vehicle? And so we can then boost that ad or boost that content so that more and more people can ultimately see it. So we spend a lot of time, and this is not just the KOL space, we're now focusing this now on the user-generated space as well. And the nice part is it sounds very complicated and tricky, and you mentioned China, we got a lot of -- actually, we got a lot of insights from China. China is probably, I don't want to say they're leading and everyone is going to follow, I think China is probably in a space where I don't know if many countries, including the U.S., will ultimately get to the level that China is at in terms of its dependency in the space. But there's a lot of learnings that we've gotten from China that we're now applying into other parts of the world around how do we start to manage the space in a far more systematic way. And as I said, it's not -- it sounds complicated, and I guess to some degree it is. We don't have that many platforms that we are looking at. If you think about like the number of apps that you all use on a daily basis, my suspicion is you're using probably 10 or less apps, right, which -- my suspicion, I could be wrong. That's typically going to be the case of most consumers. So are you on TikTok? Are you on Meta? You understand. Are you on these very targeted -- Reddit could be another good example. Are you on these platforms? And then from that, we're able to then interpret everything I just mentioned.

Great. I'm almost getting dizzy thinking about the complexity of everything. And so the next question on the marketing side, is from an organizational perspective in terms of capabilities, how do you build an organization and how does that organization interface with the rest of the company so that you can actually execute effectively on everything that you're talking about? I mean do you have marketing people on your team or do you have people from your team on the marketing teams, the brand teams? I mean how does this all actually come to be?

At the end of the day, the success or failure of any of these initiatives is the change management effort, people change management, to get it into the ecosystem. We try to take the approach of being relatively function-less as we go after these capability areas. So I would say we don't have this black box group that does work and it gets thrown over the wall for others to deploy. We actually partner with the category teams, we call it integrated brand building teams, to be able to drive all this. So far, I would say the reaction from the enterprise is incredibly positive. And the reason for it is, A, as I mentioned previously, we're internalizing a lot of work that used to be done by agencies. So people are very excited about being engaged and being part of the solution. Second, I mentioned GEO. GEO is -- generative engine optimization, is a great tool to figure out how you optimize responses back, but it's also a great tool to understand where the consumer is actually spending their time. I'll give an example. In the baby category, as moms or parents are asking questions in the wild around different types of products for their baby, diapers comes to mind, we have Pampers as a premier diaper. I would have assumed before we did this work, that more than likely these people are probably headed either into the brand sites, pampers.com, or possibly the niche sites like bump.com or Good Housekeeping. Do you know what the #1 -- one of the #1 sites was? Was Forbes. Forbes for diapers. And it was because there was an engagement going on in one of the discussion forums for diapers. And so the reason why I think that is an interesting insight for this brand, going back to your brand point, is understanding where she's spending her time is, I would argue, almost half the battle of figuring out then how to engage with her. If I'm spending all of my time on pampers.com optimizing that, yet she's over on Forbes, I have a disconnect. So part of it is that. And then what we try to then do is we bring together the whole platforms of tools that we have. So we go from understanding insights, and we can talk a little bit about how that comes to life, but then going from that, we then move quickly into the creative process, which is all the generative AI work we're doing. Same teams are involved in trying to bring this to life. Then we go into the whole adaptation to different platforms, into then the purchasing of the media, into then the full cycle back. But ultimately, I think that the organization is very excited. Now the daunting part is what was true yesterday from a technology lens and what's true tomorrow might not be one and the same, so we have to stay fairly agile in how we do it. But because we have this strong foundation, we feel very comfortable and confident that we can make those adjustments as necessary.

Great. So look, Procter has an R&D capability that is probably greater than all your competitors combined and then some. And most recently, you're rolling out one of the most impressive arrays of innovation across categories and across the world. So maybe you could talk a little bit about how your innovation process has changed with AI and maybe tie that into leveraging your incredible R&D capabilities.

Yes. And this is such a fascinating space. R&D starts with -- is probably not going to come as a surprise with the consumer. So when we talk about the 5 vectors of superiority, what we are trying to figure out is where we have the next unlock to create innovation that will improve upon the 5 vectors of superiority. One of the things that has been so incredibly impressive as I've joined P&G is P&G spends an exorbitant amount of time with the consumer. So we have over 2 million touch points each and every year with the consumer. When I say a touch point, I'm not talking just a focus group or just a panel. We have thousands of what we call connected homes where the consumer has allowed us to come into their home with IoT sensors to basically see, with quotes around it, how they're using the product. So things like we have sensors on wrists so we can see how they're washing their hair, as an example. I say see, not visual, but with the motion of the sensors. This turns into approximately about a 35 petabyte database that we have of consumers. And what we've been able to do, and I don't think any other consumer good company has this capability, is we are able to create digital twins of these consumers. Not synthetic consumers, that's an averaging of consumers. These are digital twins. And we're talking thousands upon thousands of digital twins that we can create. And why that's good for us is we're able to then test concepts with these digital twins. It's not to suggest we go from this idea of testing a concept idea with the digital twin, we go to -- we must produce it. No. But it takes this funnel down to a manageable number that we then can engage with real consumers to ultimately testing it. So it starts with this whole insight piece where we're able to take all of this information that we have and test it up against what the consumer ultimately is saying. And then from there, and we've talked about this publicly in the past, we have a very strong capability called Molecular Discovery Suite, where we're able to compress the innovation time line from what used to be years, like 5-plus years of discovery work, down to -- with the right master scientists in play, less than 6 months at times, depending on what we're trying to solve for. And this has turned into many types of innovations, whether it be innovations on the product side. So as an example, in the U.K., one of the insights that we learned was in the U.K., the consumers at the end of their dinner would take all their dirty dishes and put them in a sink full of water and let it soak overnight. And the rationale was that's the only way you're going to be able to get those clean before you put them in the dishwasher. That was the insight. What we came up with is something we call the Ferry Power Wash, in the U.S. we call it the Dawn Powerwash, which is a spray solution that sprays out. And what we've been able to do, has been tremendously successful, is we've been able to nudge the behavior for that consumer. Instead of soaking overnight, in fact, I think the slogan is "Skip the soak," and be able to take those dishes, put it directly into the dishwasher, spray with the Powerwash spray and have an amazing experience of a clean dishes and clean everything that comes out of that dishwasher. Another example is in Brazil. We had an insight of there was this worry of deodorant creating staining underneath the arms. And we were able to innovate a product very quickly that I think the slogan is "Stainless freshness," is what they call it, for Old Spice. And that too has been wildly, wildly successful. So it starts with the insight. And then from that insight, we're able to then quickly iterate through to that final product design.

Great. So to wrap things up, our research department management, and Julian Emmanuel, our strategist, have really been pushing all the analysts to really look at AI and how that's going to make a difference with the companies. And you've done a great job talking about the changing marketing landscape, how this is really going to help drive growth. But they want numbers, particularly on the cost side. And I know you're not going to give us any numbers, and it's probably impossible to do, and you wouldn't want to do it anyhow, but maybe if you could talk about the key buckets perhaps of savings. Because everything you're doing is -- costs some money to do. So how are you funding it? Maybe the key buckets of savings and maybe things that you aren't doing anymore that you used to do or things in the future, near future, that you won't be doing anymore where you can get savings and then all us analysts can kind of try to put numbers to it ourselves.

So we were speaking before our talk today, I struggle answering the question of how much do I spend on AI. Because it's almost like asking the question, I have a hammer, how much is my hammer? And then I'm going to look for nails. Instead what we try to do is we try to take an approach of, holistically, what is the capability we're trying to bring online. And then with the combination of process, people and technology, we then build this solution. But your question is a good one because there is, if you think about it, 2 key benefit areas. One is growth, so superior products, et cetera. And one is we should be able to do things more efficiently. We talked about the media example, as an example. Let me talk about a few other areas that we're focusing on that gives you -- hopefully gives you guys a reason to believe that there's some real stuff here. And ultimately, I think our performance will speak for itself at the end of the day. So at the end of the day, I don't think there's going to be such a thing as an AI-native company. I think it's just going to be a company, because everyone is going to have the AI and the companies that have adopted the best will be the ones that are outperforming in the marketplace. If we go back to those 4 toolboxes we talked about previously. So we talked about the customer consumer, we talked about the internal efficiency, supply efficiency and R&D. I think R&D we've talked about already. On that internal efficiency piece, I mentioned previously that we have a great, great, great capability with this data lake, not the sun tanning kind, but the kind we're getting all this great data in. What we are finding, and this is where I think just fact versus fiction, a lot of the generative AI press will talk about this easy button. Get the data, you get our tool or our AI capability and you're off to the races. It's not necessarily as easy as it sounds. There's this area that we call a semantic layer or ontology layer, which is the ability to have a description and a relationship of the data that is in this core data lake that allows the AI to be far, far more productive than it ever would have been in the past. And this will be, I believe, a differentiator for Procter & Gamble. So in today's world, before AI, you would take data and you create dashboards, right? That's how most people would have operated. The next evolution of that will be, if you are able to understand the relationship of the data, the AI is able to be able to understand it, you should be able to talk to your data. So instead of it being a dashboard, why not just ask how is customer X doing in this geography? And through the semantics and through the ontology, the data will be able to talk back. The next layer of progression is going to be insights that will be generated automatically. Because the AI will start to learn what's going on. And then finally, get back to this word agentic, how do I then automate a response back into the organization? So we have already been piloting -- not piloting. We've been deploying this in pockets in the organization for use cases that make the most sense. We're doing more and more of this. So as an example, we're using AI right now largely to do financial forecasting without humans touching it, as an example. There will be far more use cases as we move forward. And this is where we're spending a lot of time, whether it be in the selling organization, in the R&D organization, in product supply, et cetera. In our supply chain, it's another area. We have a big initiative on Supply Chain 3.0. We've talked about unattended operations. It's going incredibly well, where we're able to do parts of the day without people in the plants. As you would suspect, there's both a productivity point of view, but as well as I think there's actually a capability that we are able to do. The trick there was we don't necessarily just take the existing process and just put AI against it. We have to reorganize the process so that a portion of the day we can automate out of the process, where other portions of the day we still need humans in the middle to be able to do things. We have other capabilities in product supply for quality, for example. So we have tons of IoT devices on the lines where we're able to see quality concerns before they become an issue and we're able to adjust the lines very, very, very quickly. And then on the sales force side, tons of information going to the selling team so that they're able to walk into a store and be able to spend their time not trying to survey the store to see what's out of stock, what's not. Because that data is we have that data but rather talking to the store manager or store buyers to say, listen, this we're having a gap here, and I look across the neighborhood that you're in, you're underperforming other places because they have that gap filled. And we're able to see some good benefits there.

Great. Well, we've gone over a few minutes here.

Sorry.

No. No. We could go on for hours. Thank you so much. Really appreciate it. And look forward to your -- the rest of the day.

Thank you, Robert.

Thank you.