Stop Building Better Chatbots - Aniket Deosthali

0:00Hey, welcome to the CMO Whisperer Show. I'm your host, Steve Olenski. Part marketing practitioner, part ad agency veteran, part journalist, I was a writer for Forbes for 10 years. I've had so many insightful conversations over the years with business leaders, to athletes, to celebrities, to of course, CMOs. The only difference now is instead of sharing those insights through written form, I'm doing it this way. My guest today is Aniket D 'Ostali, CEO and co-founder of Advive AI. Now, before launching Advive, Aniket led conversational commerce, love that title, by the way, at Walmart, where he helped build AI-powered shopping experiences at massive scale and worked at the intersection of AI, commerce, and customer experience.

0:48Today, he's focused on helping brands move beyond the AI hype and use intelligent agents to drive real business outcomes. Now, if you've listened to this podcast for any length of time, you know I am less interested in asking whether AI is coming. That ship sailed a long time ago. What I am interested in is where it's actually creating value, where it's falling short, and what business leaders need to understand before they start chasing. Anika, welcome to the show, my friend. Hey there, Steve. Thank you for having me. Excited to be here. You are welcome. You are very welcome. All right. I know I had, I thought about the opening going, you know, whether AI is coming.

1:26If that is any surprise to anyone listening right now, you might want to consider a career change. But having said that, let's just, I want to, I want to take a few minutes at the beginning. I want you to tell my audience about Envive, right? What problem did you see that made you decide this company needed to exist in the first place? Let's start there. What was the problem Envive solves or problems? Oh, man. All right, let's go. So Envive is a CMO's best friend. And what CMOs tend to care about is two things. They want to protect their brand. So they care about safety. And because they're custodians of the brand, then they care about growing the business performance.

2:00And so Envive is a set of AI agents that creates, at scale, brand safe and revenue driving experiences for commerce brands. So we work with some of the biggest brands in the world, like Supergoop, Wine Enthusiast, Daily Harvest. We just started working with the children's toy company, Love Every. And what we do is we learn for every visitor that comes to the site, we learn how to make this experience better and higher converting for them so that they are more likely to find the right products and convert higher. And so we drive personalized shopping journeys for users at scale. And we do so in a way that the AI does all the heavy lifting.

2:37So you don't need to hire more people on your team to manage AI. What you look at is a dashboard that shows the incremental revenue that the AI is driving for you, and also a dashboard showing that the AI is fully safe, every single interaction. We work with brands that are regulated by the FDA and the FTC, and we haven't had a single issue where the AI is going off script, off rails, off the brand guidelines. Thank you, and I appreciate that, and I'm sure the audience does too, because it's a very interesting and intriguing set of offerings the company does, and it sure sounds like it could be one of the CMOs.

3:09Let's say it'll help CMOS sleep better at night. How's that? Yeah, that's the goal. That's the goal. Yeah, exactly. Okay, so before we go back, I'm going to end to that bigger outside of the world of Envive and get your thoughts on a lot of different things. But in the opening, I mentioned that you led something called conversational commerce at Walmart. So what does that mean? What does that term mean? Yeah, so conversational commerce. I had the coolest job. I believe I have today. Steve, it's the coolest job in the world. I have Walmart. I had the coolest job because I sat at the intersection of all of American shopping data, right?

3:39Like 70 % of Americans shopped at Walmart in the last two weeks. So I'm looking at that and I'm partnering with foundation model labs like OpenAI. This is back in the day. I'm kind of an AI hipster over here. So back in 2019, you know, 2018, we're using these early LLMs with Walmart shopping data to build personalized conversational experiences for shoppers. So, you know, most of AI for commerce today is a chatbot that's bolted onto a website. It demos really well. It doesn't really move the needle. At Walmart, Walmart doesn't invest in things that don't return EBITDA over three years. They invest in things that are profitable.

4:11So I had to figure out how to take all this data that Walmart sat on, along with this early-stage AI technology, and figure out ways to create personalized experiences that are profitable to the bottom line. And that's what my job at Conversational Commerce was, was to create early experiences using AI that make it for personalized site experiences that are profitable to the bottom line. And I learned a lot about how that worked. And as this technology got better, I realized that the bigger opportunity here is to serve the market by creating the infrastructure and tools, which is what Envibe is today, that creates personalized shopping experiences for retailers.

4:45Personalized shopping experiences. I just had to repeat that phrase because it's been, you know, the holy grail, right? To actually have that personalized, just experience, let alone a shopping experience. Yes. Yes. So two things are happening at once, right? Like The consumers are starting to shop through AI and the storefront itself is getting more intelligent. And CMOs are reacting to the first thing, but they're missing the bigger opportunity, which is the second, which is that like in the old world, you know, the last 20 years of e-commerce and AI, like e-commerce technology, you build a page, you have your team run an A-B test by hand.

5:18It's frozen until someone looks at the analytics and realizes that, oh, this, you know, landing page converted better. That's the ceiling that I think that every CMO is facing and has been facing. That's what my teams at Walmart were struggling with. because we're running a Walmart scale, this is most of America shopping. We're running like thousands of A-B tests every month. And what I realized is that what we need to build is agents that run their own tests and learn faster to figure out what it is that drives revenue growth at scale. And that's what Envibe is. Okay. So for seemingly forever, we've, and I say we marketers, we train consumers to type keywords into a search box, right?

5:52Now you say, and you believe that, that behavior is changing. What's actually happening? Yeah. So from a consumer perspective, of course, consumers are using, whether it's a Google search, just how Google has that like AI suggested prompts, or they're using ChatGPT or perplexity, they're using AI in some capacity to find the right products. But then they're not transacting with the AI, they're transacting on your site, or they're coming in through paid social channels, right? Like a lot of our customers, 30, 40, 50 % of their traffic comes through Instagram ads, right? So the question is, how do we learn from what the user's intent is, whether they're coming in through ChatGPT or they're coming in through a meta ad to align to that so they're more likely to transact at that moment of truth?

6:36And then remember, you're going to have different consumer segments, right? Like our average customer has five, six, seven, a dozen different personas or segments they're targeting. How do we personalize to each segment so that we can see the revenue upside for each segment? Yeah. It's the shopping, personalized shopping journey we talked about, right? and you're able to provide that. But let's take a second and look ahead. How do you think that journey is going to look in say three to five years? Yeah, what a great question. So the way that I think that all things kind of trend, I'm going to get a little philosophical here.

7:06I think everything trends towards truth. And so like that's what we're already seeing is that there's no discernible difference now between UGC and ads. A lot of the best performing UGC ads are basically use generated content, you know, which are basically used as ads. So I think what's going to happen is As things get more personalized, ChatGPT or whatever your element choice is, we'll learn more and more about you. Then from there, the ability to take that and then take that context, personalization about context, and it's about storytelling and bridge that into really immersive site experiences is where the future is going.

7:39What I don't see in that, what's missing from that is that there's a lot of venture capitalists out there in the valley who are viewing the world that the future is AI to shopping on your behalf. I think that outside of just soap and, you know, toilet paper, I think that most people like shopping and they don't want to let go of that control. So I think what's going to be the inevitable future that we all have to prepare for as growth marketers is like, how do we make sure that I can talk one to one with each person that's coming to the site? And that's going to be about taking the understanding who I'm talking to and what's going on in their mind at the point they're coming to the site and are open to a transaction.

8:11How do I engage them meaningfully? That's what the next three to five years is going to be about. And the best brands, if you look at like Grooms, who just had a fantastic exit, the reason they've been able to do that is they've made some early investments in scaling up how they talk to their consumers in ways that are personalized and immersive. Yeah, it's interesting when you talked about the toilet papers and the paper towel, like those staples, right? Those, I agree, it could be just put it on order, you know, replenish, right? But if I'm going to go buy a pair of shoes or a pair of pants or whatever, I'm going to talk to AI first, but I'm not going to ask it to make the purchase.

8:49I still want to do that. Is that kind of where you're going with that at a high level? That's exactly right. Like I think humans outside of those staples, I call them low consideration purchases because you don't want to think about them, right? You just want, you know, the rest, like we're not going to have AI shopping on our behalf. We're still going to want to discover brands. It's still going to be human process. So as long as humans are making decisions, the things that great marketing will always win, right? Like how you market the bots and AI versus humans is very different. And we as humans have spent a long time learning what gets humans to buy things.

9:21the question now is how do you scale that on a one-to-one basis like me and you and it's like you know you and me steve right like steve you as a man have very unique preferences and needs very different for me what cmos need to figure out now is how do we the next three years user journeys but understanding about what we can about steve so that we can make sure that we're as compelling to steve as can be at that moment of truth when that when steve is ready to transact i know in the past you've talked about and we might have talked about this off air but about brands losing visibility and even control over how they're represented inside AI systems.

9:57And that's a little unsettling, right? At least to me, and I would imagine my audience, can you elaborate? What do you mean by that? Yeah. Yeah, it's really interesting, right? Like how you, just like with SEO, how you're represented is, it has never been fully in your control as any business. But with AI, it's a whole different level because with ChatGPT, Gemini, like the brand, you can get a very, because these answer engines and these AI systems are forming an opinion and a perspective based off the available data, it can be even more challenging because it's not just bumping, oh, are you at this result or this result?

10:33It's now, hey, Spanx is good for this, but not for this. It's a very opinionated data set, right? And so the thing to keep in mind is that there's four key ways that ChatGPT or Gemini, you know, basically form a perspective on the brand is based off of Reddit, YouTube, publisher networks, and then your own content. And of those things, you can control your own content and your own site experience. You can kind of control YouTube in terms of the content you push to YouTube. But besides that, it's really tough to persuade or influence other systems, which is why it's important to have a great product and then to also have some great content on your own site to ensure that you're able to effectively control what you can control in this world where things are so seemingly out of control.

11:17So if I'm a CMO today, or you're talking to CMOs, what's the biggest blind spot? I probably have around AI-powered commerce. Yeah, so AI-powered commerce is a lot of things. It does a lot of things. So I like to first break that down towards like acquisition, like the user journey. So AI-powered commerce between acquisition, browse and consideration, and also checkout, but I'll actually go check out for now. There's different things for CMOs to consider there. at the acquisition level for driving more qualified traffic to your business. The number one thing is going to be the investments you're making in GEO, specifically your own site experience and understanding your first party data.

11:55So a lot of ranking better in GEO comes down to better first party data collection and then creating like long tail content against those queries. I know this is a bit tactical, so I have to go up a level here, but I like to keep things as like in the weeds is useful so people can walk away with some action items here. So making sure that, remember, the average LLM query is like 37 words, whereas the average SEO search query is like three or four. So really understanding the long tail of things that matter for your brand and your business starts with collecting first-party data. So that's what matters at acquisition.

12:28For browse and consideration, it's a bit different. At that point, CMOs need to think a lot more about on-site personalization, especially if you're an e-commerce brand and you're DTC. if you're selling anything on the internet through your site in some capacity, that's where personalization and having systems that basically autonomously create off experiences and learn from them is really important because most businesses do not have, you can use like tools like Replo, landing page generation tools, or even Cloud Code. But the challenge is that if you're doing any amount of scale, like 25 million plus in GMV, at that scale, you need to have, you just have too much going on to effectively use a cloud code type system to just plug in.

13:09There's just a lot more complexity to the different campaigns you're running. So at that point, for browsing consideration, for AI-powered commerce, CEMOS need to think about finding a partner that can help them scale out their on-site experiences. Yeah. Merchandising, that's another, I know, big topic for you. And specifically, I think you've called this, or I've saw it somewhere, the next evolution of merchandising. And what I'm talking about is you've described AI-driven merchandising as the next version of digital merchandising? Help us understand what that looks like. Yeah. So AI-powered merchandising is absolutely, that's what I'm actually referring to when I say the browsing consideration aspects of the buying journey.

13:45So like I think of merchandising, you know, back in my Jet days and Bonobos days, you know, I've worked very closely with the site merchandising teams. It's really merchandising is all about how do you talk about, how do you change the, just like it's in a store, right? Like in a store, merchandising is all about how you, where you put products, at what height, you know, the merchants at Walmart, where we're very OCD about what shelf something's on. In the same capacity, site merchandising is all about how you're changing the site to make it compelling for people to buy stuff, right? What AI is really good at is testing and learning way faster than humans can, right?

14:18So teams of agents can run tests, form-hop the hypotheses and execute them like 100x faster than your team can today. And so what CMOs need to start doing is to start experimenting into tools that enable AI-powered merchandising. because if you are doing north of 25 million gmd you're definitely looking at some material upside there if you as you start to get your your feet wet in this in that in that technology and tool set so so talk about what are the brands getting right today that others aren't what are the brands getting right today look like i take a lot of inspiration from like grunes right like grunes there's a couple more uh what's the i i'm honestly i'm blanking on the name but there's the there's the other one that is more like athlete focused supplement brand.

15:03But what they're doing right is they are focused on like the commerce space is one where you need to be working backwards from that customer, from the revenue outcome, right? And choosing a specific user journey. So like take Klaviyo, right? Or take like an email retargeting journey or take like a funnel, right? Like choose a funnel and then start thinking about like, how do I scale this up to start working. And what Grooms did a fantastic job of was they focused on, they had a compelling product, but then they're running lots of ads. They realized that new user conversion is the problem that we're going to focus on.

15:38And they started to experiment with creating super immersive experiences that are really compelling to work there. And then they crushed it from there. So like, that's where we need to focus. Oh, sorry, that's where CMOs need to focus is like choosing one problem to solve, whether it's like this email retention pace with this new user funnel, And then getting creative about where there's going to be alpha from an AI system coming in, automating that, the journey towards revenue growth. That's really interesting. But I want to look at the inner dynamics for a second between the relations between some teams like marketing, commerce, product teams.

16:12What does or how does this, I should say, affect the relationship between teams? Yeah. So things are kind of converging, right? And so I've seen situations where marketers are not doing engineering work. I've seen situations where an engineer just sent me, like they're owning acquisition numbers. And so I think what matters is that you have someone that's owning the business outcome and then is making pragmatic decisions about what is the tool set I need to actualize that outcome. And there's certain cases where it makes sense to build and there's other cases where it makes sense to buy or partner with the right tool.

16:43And so that's one of the things that the smaller and mid-market brands are much more agile and nimble as they typically are to navigate because the org structures are converging, which makes things a little bit more challenging if you're a more scaled out brand and you have more fixed org structures. Okay, so as AI, commerce, CX continue to converge, if you will, does a legacy brand have an advantage over a challenger brand or vice versa? Or does it really case by case as we move into the next one, three, five, seven years? Yeah, it's so interesting, right? Like it is, I continue to believe that a brand is about an emotional connection.

17:19And then of course it depends, right? Like if you have like, think of like, you know, Sling or Comcast as businesses, you have some structural assets there too, right? Like, which is like, it's not easy to deliver cable to people or TV or any kind of entertainment. But like when I think about commerce brands, right? Like I think that having a differentiated product and a real capital B brand there is still like super defensible. But it is easier than ever for someone to take like, look at what Grunz did, right? Like they took athletic greens, they put it into a gummy and then they were they built a beautiful brand, which is super easy to do with AI now to create like all the brand assets.

17:54And they, you know, they went from zero to a big exit in a fairly short amount of time. So so I think that what that means there is I don't think that incumbent brands are screwed in any capacity. I think that they need to be leaner. They need to build AI into their systems to run leaner and act like challenger brands. And challenger brands have the lowest barriers to entry that they've ever had. And then the focus becomes how do you build a great product and have that narrative. Interesting. So each legacy and challenger have their own challenges, of course, which is true in anything. And this is no different when it comes to AI and commerce and CX is basically what you're saying.

18:30Yeah. Right. Like I think put differently, I think that Spanx is at the Kleenex of shapewear. So like these iconic brands will continue to be iconic as long as they're managed effectively. And I think that they need to hire like chief AI transformation officers, right? Like to basically help them retain their brand credibility. And also I think that challengers have the lowest barriers to entry they've ever had, where you can spin up and get something moving fairly quickly, especially if you are a category creator. Like Grunz was in the gummy like supplement. There's not so many people in that space, right?

18:59Like so they took something like they didn't get into powders. They didn't get to the AG1 space. They took AG1, put in a different kind of like form factor. And then they went and owned that category, right? So like challengers who are new category creators have an especially awesome advantage to come in and win. And Groons is just one example. There's a few more light Groons that we can draw on. And that's my framework for how things will play out. Well, I got you and thank you. And now it's time to have some fun if you're ready for the random five. If you've heard any of my episodes. I thought I was already having fun.

19:28I guess we can have more fun. How's that? This is when the audience gets to know Anakin as a person. Are you ready? Let's do it. when you're completely unplugged from work and i heard that does happen from time you know what are you usually doing what is i can like to do away from yeah i am usually in the ring you'll find me i like to box so you'll you'll find me you'll find me hitting some pads or getting hit so that's a that's that's one of my passions is uh is boxing so or playing with my dog but not boxing your dog no my dog would be but it would definitely beat me up so i only fight humans unfortunately good to know okay yeah what's something you're not you're just not good at and you've just finally stopped trying to pretend you are man honestly being a ceo there's a lot of stuff that counsel wants me to say in a specific way and i just cannot i'm not i'm not good at i'm not good at being a professional ceo probably if you define that yeah corporate speak there you go yeah like oh that kind of thing i gotta be authentic man that's why that's why you and i get along and we're going to do something together.

20:30We know it. Yeah, 100%. What's the best piece of career advice that you've ignored? You know, probably, this is a really good question, best piece of career advice that I've ignored. You know, I think that it's probably like, I started my career in consulting and it's kind of like, I put my, as a lot of people do, I could have basically stayed in and stayed on a very safe track, like probably just like following the tracks that are available. And I think that it's like really good career advice because I think that predictability is important to a lot of people. So like being able to stay on that predictable path and know like, hey, if I stay this or make these moves, I'm here.

21:06But I'm also grateful I ignored it because I decided, just to give you some context, so this makes sense, but like I decided very early on in my career, like when I was 24, that I want to be at the intersection of machine learning and commerce. And I was like, I'm just going to go deep on this stuff. So like that led me to do some really weird decisions, where like I had an opportunity to be, I had a product of a, I won't say the company, but very large company for billions of dollars. And I turned that down to go to Walmart. And at the time, everyone was like, what are you doing? And I was like, I kind of decided this track when the kind of like the career ladder move would have been to go there, get that name brand on my resume and all this stuff.

21:38And that's been really dividends for me because I just figured that like, hey, this will, just getting knowledge and expertise in this space would be really valuable to the world. And that's so far worked out. Yeah. What's one thing people assume about you that's just not right? It's just wrong? man i wish uh i want to be more self-aware to know what people assume about me but uh you know probably that that you know ceos right like now i'm not gonna try to you know this is a thankless job so i'm not trying to like like be out here looking out for ceos right i feel like i probably just thought like hey like this dude just like is focused on shareholder value all this kind of stuff and it's like yes of course like that's my core responsibility but like what i really care about you know the day is like i think that you only get one shot at life and life is about doing like cool things with cool people.

22:21And so just like that's just my outlook. I'm just a very, I have a very, I can come across very intense, but at the end of the day, it's like, I think that we're all humans. And I think that people don't understand just how casual I am about things. I'm just like, look, like people mess up. It's part of the experience. I love that. And I hear that a lot from not just CEOs, but just C-levels as well. And there's a perception and maybe it's when you're at a younger age is that, oh, I can't talk to Hanukkah. He's the CEO. Or, you know, I can't say hi to him or something. You start making up these scenarios in your head.

22:49Meanwhile, you're going, I'm the most approachable guy ever, but just because there's three letters in my title, you know? Yeah, there you go. Actually, Steve, you did a better job of articulating than I did there. It's like, I think that's exactly right, right? Like, as a CEO, like, I actually think that I consider everyone in my company, like, family, like, genuinely. Like, I mean, like, I really care about people, not just my company, but, like, just generally speaking. And so it's interesting, because in this role, or even being an executive, like, you've been an executive, right? Like that's one of the biggest places, but like you, you get reduced to like the set of decisions you have to make and you kind of lose that humanity that is actually, I think, core to all of us.

23:25Yeah. Okay. Last question. What's, take us to your, your routine when you get up in the morning. Is it a phone, email? What's, what's Aniket's daily routine in the morning? Yeah. So let's say I'm up around five. I'm going to immediately go for a walk just to, just to get my head straight. and then I'm going to go and the first four hours of my day are spent, you know, really promoting the business, right? Like just getting the word out there, getting the good word out there to the people. And then from there, the next four hours of my day are operational in nature, you know, checking in with the team, see how things are going.