Hello and welcome to the retail podcast. Now I've had the pleasure of seeing Keith actually was I think last year at Shop Talk Spring when Microsoft had a, a room to the one of the sides and you were one of the, the, the speakers there, Keith. And for Full disclosure, I have known Keith from my previous life at Microsoft, but that shouldn't stop us talking about some of the awesome things that I know Keith has been travelling the world telling people what
about what Microsoft is doing. But Keith, for those people who don't have that intimate knowledge of who you are and what do you do, How about you tell us who you are and what do you do at Microsoft? Of course. And first of all, thank you for Alex for having me on the retail podcast. So excited to be here. I grew up in retail. I spent 27 years working for Gap. It's in my blood. So I love having retail conversations.
But for the rest of you, I am the Vice President of the worldwide retail and consumer goods industry team at Microsoft and our role is to work with our top retail and consumer goods customers globally. We are all ex industry folks. We I always call us the great translators of customer business problems into ways that Microsoft technology and Microsoft partners can help solve those problems. So thanks for having. Yeah, I know.
And I and I think it's a testament to how well Microsoft has grown within the retail community, how your team has basically been able to address those major issues and what a time it is for those major issues. So I mean, I'm, I'm really curious. You, you meet executives day in, day out. I know. I'm surprised you're at home, to be honest, right? You're not. It's, I mean, the airport somewhere. Yeah. What's well, what's what's the thing that's top of mind right now?
What are what are people coming to you? And talking to you about or wanting to talk to you about. I mean, it's really about a gentic AII mean the the pace at which this technology is changing and the importance that executives realise that it can have on their business is probably unparalleled than any other technology that I've ever
seen. So they really want to understand how can they uniquely adopt these capabilities to reduce cost, to give better customer experiences, to build a more agile supply chain, and honestly make use of the incredible value of data that they sit upon. And retailers have an incredible amount of data. And for years they generate data, but they haven't been able
to really unlock the value. So with generative AI, agentic AI, they're seeing finally this uniqueness of how they can take that competitive advantage and put it to work to change and transform their organisations. If you don't mind, if we do like a quick one O 1 on what a gentic is, because I think sometimes I mean, it was the IT was sort of the on on stage at shop talk this year, shop talk spring. It was like, you know, a gentic is just from from the person who's in data and AI.
You you tell us what is a gentic. AI yeah, yeah. And I'll start back from when generative A kind of first came on, you know, came in the market with GPT, which was this natural language interface, which which you could ask questions, generate content. And that's that's where we started. And it taught us all how to have the power of the prompt, right. The output was only as good as the input. And we all had to learn how to do these inputs.
So at Microsoft we took that capability and started launching Co pilots and Co pilots were the ways you could access your data. It's the way you could help generate that first draught of an e-mail. And that's really where we started. Now you layer on a gentech and the good news is we believe Co pilot is the UI for all AI. So the way we will communicate with agent with agents in the future is the same way we've been communicating in the last year and a half by using prompts.
But agents now can do more than where we started. So the first thing is more about retrieval, going out and finding a piece of information and bringing that back to you. You're still directing it using natural language. Then it will move to tasks, so performing a task on your behalf. So I think about the marketing function and every campaign starts with a brief and we used to use Co pilots to write that first draught of the brief.
Now we can have an agent write the first draught of the brief, and once we build confidence in that agent, over time we will see those agents act autonomously and start to generate briefs on their own based on the insights the agents
are seeing in the business. And that is really the unlock of value because now the human intervention is still always going to be there, but it will be significantly less than it is today, which means we as humans, we'll be able to do higher level thinking, think more strategically, and take a lot of that kind of lower level work, like brief generation and business insight, you know, analysis.
The agents will be doing that on our behalf and surfacing those insights and recommendations to the humans. And we could just say, Yep, keep moving forward. So that is really going to change the game in terms of humans and agents working alongside of each other. I know Satya, maybe one month or two months ago, he gave an interview where he said it was the death death. I can't remember the death of Sass or Sass is gone or I don't know. I'm just curious.
Was that just, was that like a sound bite someone took or is that genuinely how Microsoft is actually seeing it because you have so many products underneath your agentic just from a Microsoft perspective, if I'm the Microsoft customer? And in the spirit of that comment is around if you think about the number of business applications that we use to do
our jobs everyday. So if you are in a marketing function, you probably have some data that's sitting in Excel, you have some data that's sitting in some type of web analytics tool. You have other data sets that are sitting in in your advertising tools. And all of those are individual business applications that you're going in and out of every day to gather data. So you can do some analysis potentially in a fourth tool like Excel. Yeah.
And what we're, what we believe is that agents will now be the, the systems going in and interacting with that. So we don't have to do that any longer. And we'll simply use natural language as our communication and UI through a Co pilot. So you no longer have to go in and out of all of these different business applications to get your insights. The agents will be doing that on our behalf and we'll just be asking questions. And it doesn't matter where the data sits any longer, right?
It doesn't matter if it's in this tool or that tool, as long as you trust the data, as long as the data is secure and compliant, then we'll let the agents do that work on our behalf. So I think the spear of that is more about natural language will be the UI throughout your Co pilot, which was how we'll start to interact with disparate data sources across the entire company. Yeah, I mean we, I mean that makes perfect sense. So thank you for. Yeah, of course.
Clarifying that, I remember from days gone by, we used to talk about retail operating at 2 speeds in terms of, you know, almost people that have structured the data, people who have spent time, energy, money and effort over over the years getting some form of structured data. And then you've got this other sort of wheelhouse of retailers that haven't done anything and everything lives in silos.
I'm just curious. I I guess this this question breaks into a couple of points and I'd love to know if I'm a Microsoft customer and I sit in in those two worlds, does a gentic AI actually help me or no, sorry, you still need to get your data into some order. That's the first question. And then if I'm not a Microsoft customer, but I'm, you know, I'm saying here, listening, going, holy cow, this really sounds like, you know, I I'll be with the dinosaurs.
If I'm not going ahead, I'm doing, you know, Where did you start? What does it look like? Well, I love. It we're having a data conversation, because it is, I always say every generative AI conversation always leads to a data conversation. You can't have one with the other, right? You can't. And you, you said it, Alex, like retailers traditionally have siloed data. It's just how they've been built.
You know, many started as brick and mortar and built entire organisation and tech stack around that. Then they launched e-commerce and that was a different data set. Maybe they went into a different model or wholesale, whatever. But the disparate data silos have been a challenge in the industry and for years. You're right, many retailers have been working to bring these this data together. It's hard, it's expensive, it's slow. It's really a process.
And what we're seeing now is we've launched the data platform that Microsoft called Fabric that is less about breaking down the silos. It's leave the data where it is. We can just help you extract what you need. Well, so you can build that first AI use case. And I'd say what we're seeing in the industry is this like I've got to get my data house in order before I can even start. And yes, that's true. But we have ways to help you
fast track that. And we're, that's where we're spending a lot of time with our customers today is let's pick a use case. Let's look at the data sources that drive that particular use case. Let's unlock those first. Let's feel good about the data. Let's make sure it's secure and compliant, and then let's launch a use case. And I think that has been actually one of the barriers we've seen.
We've seen a lot of our customers launch a lot of Pocs last year, but they've kind of got stuck in this pilot purgatory where they have been able to move to scale. And we really want to help them take those use cases to scale. And the data unlock is is key to that. I mean that that's a wonderful segue in terms of obviously we can dwell on the whole data and AI element, but I can on the second speed where people are moving at blistering speeds.
What are some of the use cases that have really sort of impressed you? Yeah, on your travels around the world. Well, you know, there's two that we're seeing a lot of traction and there's many, but I'll, I'll talk about two. And I like to talk about these two in conjunction because one of my roles many, many years ago, that gap was when I first started, I was a sales associate. It was my first job.
I was in high school and I was folding jeans and I was, you know, I had to learn everything about our products. We had this printed manual and we studied it every season. So I could talk about features and benefits, but we didn't sell very much. It was jeans and T-shirts and
sweaters. It was relatively easy, but I think about the complex nature of the sales associate today in retail and some of the products that customers and offer and how do you learn everything about all your products in a confident way where you can just help a customer on the sales floor. And I've always said, you know, retail brick and mortar is very
service rich but very data poor. And this is where agentic AI is going to help bring the necessary data into the fingertips to the associate at the moment they need it, which is the moment they that matters with that customer. And I'll I'll tell you a great customer story with me to mark Saturn. If you don't know them, they're yes, I'll see 3 Taylor in Germany. I'm in that carry a very broad range of a highly complex and very similar products across many brands.
And they've launched this my buddy in in my ear Co pilot that is a on demand basically phone a friend to ask a product question and have it answered in your ear while you're working live with the customer. And it really is changing the game of how they can answer questions about think about, you know, big screen televisions. I have 5 of them that look very similar. They're different brands. The price point's very similar. How do I explain to a customer
what the differences are? So yeah, my buddy has been a way to help bring that information into the front line worker at the moment they need it. So bringing that that data along with the service, now I'm going to take you to the e-commerce world. And that was a role I also had at gaps like I have to run the e-commerce business for a while. And you kind of have the opposite problem. It's actually very data rich, right? Data is just a click away on the web, but it's very service poor.
There's no one there. You're you're kind of self-guided, right? And e-commerce hasn't even changed very much in 20 years. You've got your navigation across the top, some categories down the left, your search box, but that's kind of it. So conversational commerce is changing how customers solve problems. So if you're trying to throw a World Cup watch party for 10 friends, what should I buy? Well, you might need drinks, you might need snacks, you might need a large screen television
and a great sound system. So companies like Walmart are bringing that type of conversational commerce experience to the web. And what they're seeing is it's their customers are moving from this kind of scroll based shopping to goal based shopping. And we're seeing other brands launch this and we're seeing increases in basket size.
We're seeing speed through checkout, which are all great signal that consumers are engaging on this and they're finally able to shop by solving their real problem versus kind of searching for things. So those are two of my favourites and that we're seeing today. And and so just just to sort of understand that in terms of the Walmart use case that you were saying you were talking about there. So just so I've got a clear in my head, I'm I'm having a BBQ. Is that me?
Instead of typing with the e-commerce page or site I'm talking to, it is that is that it's. Oh, can you just. Yeah, it's right. You're actually seeing not only, and now they haven't launched this yet, but we're actually seeing multimodal where you're saying, I've got this a hiking shoe. I, I want something just like it. And you're holding it up to the camera and have an agent behind that camera who's recognising that and comparing that to your
product catalogue. So it's text, it's voice. And very soon you'll start to see multimodal commerce where you as a user can interact with different ways that are most comfortable to you and helpful. So where Walmart is starting right now is they've added more conversation to their search, but we'll eventually move to a full conversational commerce
experience. Other examples are Bath and Body Works in the US has launched full conversational commerce where you can literally ask a natural language question around I'm, I'm looking for a gift from Mother's Day. What ideas do you have? I mean something like ask a sales associate in the physical store.
I think, I mean, that sounds a little bit like because I saw the CEO founder of Perplexity sort of saying that's what they're trying to build or building or training their models to do. But obviously as a retailer, I want to own and have as much control of that. Why would I want some? I mean, obviously I'll, I'll be open to all channels, but why would I want someone else to have that data on, on my business? So that makes so much sense because I I, yeah, I remember.
The point on data is important, one, because what we haven't talked about in the case of conversational commerce, imagine if you're a fashion retailer and you offer that experience and a woman comes and types in. I'm going to a wedding in Greece in June. Can you suggest an outfit? Yeah, of. Course. And now as a marketer, I now know it's, you know, it's early May and next month you're going to Greece in June. I've got some real context to start marketing to you about.
Well, here are some great items you should buy for sightseeing while you're in Greece. And have you thought about a swimsuit for the beach in addition to the out that we recommended for the wedding you're going to be attending? So you could start to see how the data capture of conversational commerce is very strategic for retailers because I'm getting more context about you as a shopper, I could hand that back to your customer data profile so I can use that in the
future. So it's really opening up this opportunity for true personalization that we haven't seen before. And I think that's another unlocked. It's not just the the the existing customer experience that we're going to see, but it's this incredible data capture that's going to come out of it. Yeah, and it makes so much sense. What do you see as the blocker? What's the barrier? If you were going to do a CHEAT SHEET, you know what? What's holding people back?
I think there's a few things. I think 1 is I talked about, we talked about data is like how do I start to think about my data strategically? And, and it was a this, this isn't the CI OS problem. This is the CEO and the executive leadership team is a retailer's data is very strategic because it's incredible about their customer, about their products. It's unique to them. So there's that. There's the second part is how do I start to get off and running with with a use case?
And part of that is assigning some business value. So we do spend a lot of time with our customers helping them prioritise use cases around business values. They know which ones that are going to, you know, offer the most cost savings or revenue uplift, whatever the KPIs are they're trying to drive. So I think that's the second 1/3 is organizationally, you know, it's these are going to change
the way we work, right, right. And if you, if you've had a chance to use a Co pilot everyday, I know I do in my job, I'd have to change the way I work everyday because I now have a Co pilot that I can go to first to ask questions and summarise my emails and tell me what I missed when I was out last week on vacation. That's very different way of working than I used to. So yeah, it's going to change the way teams work together. It's going to change the way
functions work. So I think that kind of cultural change and organisational change is also another one. It touches on in terms of you were talking about this bimodal or, you know, trimodal in terms of voice, text and image.
Yeah. In terms of when you look at the future then I don't know if you talk about quantum computing and the chip, but forgetting that for one second, practically when you then think about the future of retail, it feels like you're describing this sort of multi sensory e-commerce world, right. Is that, is that where we're going? But I think it's a world where shoppers are finally going to get the level of service in stores that they that they
crave. And I think as frontline workers are finally going to get the data that they want at the moment that matters. It's going to help their jobs be more enjoyable. It's going to make them feel more confident when they work with customers. So I think we're going to start to really see those things come together in ways that keep associates happy in their jobs that make cut. Edwards and associates.
You think today, when a customer and a sales associate has a conversation that is dark data, it goes away and there's some really valuable information in there. So I think in the future, when you think about conversational commerce capabilities, we'll capture that data. And finally, retailers will be able to deliver on that promise of personalization that they've been really going after for years. So I'm excited because I think as a shopper, our experiences are going to be elevated.
It's just going to happen. As a retailer, our ability to learn more about you and use that with your permission to make your future shopping experiences better is finally
going to happen. And, and that frontline worker, you know, who's been so underserved for so many years as a persona is going to be happy doing their jobs because they're going to have these incredible digital tools in their hands or in their ear that are going to bring that information that they need and just help them get through their day, just like we
are every day in the office. So I, I'm excited by, you know, about how it really is going to raise the ceiling for everybody, but also raise your floor for everybody. So I think that's really exciting.
Do you think I mean you, you touched on a really important point and I don't know how many read because just going back to the perplexity or, or search, so many, I mean, when Alexa came out, people were like, oh, this is fantastic until they, and a couple of US retailers, large ones realised that all of those search terms invoice, they had no idea. They, they had absolutely 0 idea what people were searching for.
And it feels like people or retailers haven't woke up to the fact that, yes, there's lots of great platforms out there, but that data isn't yours. You're not going to be able to get context on that. And, and I remember from years gone by, Microsoft was really sort of it's your data, you, you, you get access to all of that. We don't do anything. We just give you the cable. Is that still the case? Is that like because it's such a fundamental point? Yeah, definitely.
Like we're not training our models at 1 customer and taking them to another. I always say each instance of AI is kind of like your own child that you're raising with your data. Got you. And that stays? But it's a really important point. I think it's so important. Yeah, we, we definitely obviously having like, like retailers must have a trusting relationship with their customers. We need to do the save.
So when we talk about our AI that runs in Azure, that is being utilised by retailers to build conversational commerce experiences or build frontline worker experiences, that is all trained on the retailer's data. That is their unique first party strategic advantage. So obviously that stays there. So every platform is unique. Just like today when a customer runs on Azure that that data is not commingled with any other customers that run on Azure. It's very walled off that way.
It's critical. Yeah, no, OK, that's, it's good to hear because I think that's that's something that people are sleepwalking into, if I'm honest, you know, well on that point. Though I would say what is really becoming critical security, because if you think about democratising data and putting it into the hands of folks that might not normally have it, but you're doing it because it's going to bring value, it's going to save money and time, it's going to allow for new experiences.
How you secure that data is now more critical than ever because obviously we know when when there's a trust break between a data breach at a retailer takes many years for trust to be built back. So we always talk a lot about security as well, or I always say get us your chief, get your your chief information security officer involved early and often when you're doing AI use cases because you want to make sure that everything you do is still within the guide.
The guise of like secure and responsible AI, which is the other bookend of that is using this capability in a responsible way that make it your customers feel confident it doesn't violate privacy laws. So all of that is in critical to all of these use cases as well. It's foundational, honestly. I think, well, I mean, look, we've had so many public cases recently in the UK with two major retailers having outages. I mean, there's, there's it's like who hasn't had a security
or cyber event? I get, I guess it gets overlooked because people sort of almost think that the Ciso's taking care of it where, you know, it's a culture thing as well, right? You need everyone on the same page. Yeah, but in this world of privacy, GDPR and everything else, there's a whole another, you know, how are you protecting the critical data that you're gathering on, on your on your customers? No one wants to know that their their AI data set has been
stolen. I think that's that's another level from your credit card. But again, that's like we're getting into I think that's a little bit further down the line that people, people need to be aware of it in, in terms of where you are going to be focused on for the rest of the year. Is that due? Because obviously Microsoft again operates in, you know, four or five different product areas, right?
Cloud security teams, what a workforce in playing what, what are the things that you're going to be like? How is the rest of the 2025 going to be for you? I'd say we're seeing this adoption of a Gentek AI mean this is going to, as I said earlier is helping our customers think about natural places to adopt. And I think, you know, marketing content creation is a great example. I talked about that earlier.
You know, I think about business analysts, the merchandising and the, and the inventory management sides of the world. I remember, you know, my days at Gap, every Monday morning, our inventory folks would come in, our merchants would come in, They'd spent 4 hours kind of diagnosing the business. And that was about 20 people doing that, right? So there was like 80 hours, 80 human hours every Monday
diagnosing the business. And then in the afternoon, my leadership team and I, there's five of us, we would decide what we were going to do about the business in the course of an hour. And I think about the imbalance, 80 hours spending, diagnosing 5 hours spending making strategic business decisions. That's an imbalance, right? We we need to unlock more of that 80 and shift it to more strategic thinking. And that is where a gentek is
going to make a real impact. You know, maybe those eight out those eight, you know, 80 hours become 20 hours because over the weekend agents have been diagnosing the business. So when when the merchandising team comes in and Monday morning the insights are ready, some recommendations are already teed up and maybe even in the future of autonomous, the first draught of the marketing brief has already been written on the promo that needs to run as a result of last week's business.
Like imagine that happening and being kind of ready on the desktop Monday morning. And that is really game changing when you think about running your, your, your retail business. Yeah, I mean, you, you, you touch on a this isn't, this isn't going to be a self promotion about a book, But I, I like many people, I've written a book about the future of retail, but I talk about the store being
your signal generator, right? So all of the signals that the store's generating that you then need to respond to, whether that's a, a TikTok trend that has been, you know, going up and up and up. And if you don't tap into it or it's a Netflix movie that has, I don't know, blown people away and people want that jumper in that colour. Those signals are currently, you know, again, we're going to that people are trying to sort of chip away at those types of
problems. Yep. But the store is a is this is a data centre. There's petrol bytes of data that's going to come out of that store and and how do you tap into it? So I think the signal story is is brilliant because it's also a data problem, right? So if you know they're right, traditional store signal sits in rows and columns, right? That's it. A TikTok video is an unstructured data source. An Instagram feed is an
unstructured data source. So something that's trending on a social platform is mostly unstructured data. So really it's how do you take both? How do you take the traditional structured data from the store? And by the way, you multiply that by miscue count, by region, by store, right, by how many countries you operate in, plus
all the unstructured data. And how do you take all of that signal and turn that into some recommendations on what you should do next, whether that's what you should be buying next year because there's a trend that's emerging that you want to make sure you capture on or whether it's responding to out of stocks in a location based on a significant change in weather and all of that signal and resiliency is where generative AI, a gentic AI is going to play a role.
It's it's you just can't throw enough humans at it. It's just impossible. It's costly, it's slow, it's filled with bias. Like this is where I think we're going to see such that this really step function changes. Retailers will finally be able to kind of take all that data signal back in even from your commerce, like all that data exhaust that comes from all the clicks and the movement and really make sense of it in real time and start to act on it in
real time. And that's going to be really incredible to watch. Where, where, where is is mine? Where's Mike? I know Microsoft has the biggest presence at NRF, which is also, I don't know are you anywhere else throughout the year is Microsoft, are you at any other events or is there anywhere else people? Could still definitely it's one. More plan have a. So retail NRF in New York is definitely a big one. NRF Asia, we have presence there in June, say cool.
Those who are going to be there with definitely be there talking. Stage Yeah. And then we also on the consumer goods side of the business, we go to the consumer goods forum every year and present. And then this year, I'll be at camp, there's a lot of our customers, marketers go to Can and we have some events planned around that, specifically around how generated by Agentic AI is transforming the marketing room of the house.
That's fantastic. Well, Keith, thank you so much for giving up some time to speak to us. It's probably early in the morning for you there, right? It's not so bad. It's not so bad. That's wonderful. Not bad at all. And thank you Alex. It's just a pleasure always to chat with you. We've I've always liked chatting with you. We've never had a problem talking. So in six months, let's do it again because a lot will change as we know and we'll we can give
an update. Well, hopefully, maybe maybe joking aside at NRF 2026 with. Less 2. Yeah, wonderful. Thank you.