Retailgentic | Consumer Behavior & Retail Trend

In this episode, we sit down with Michele Fisher (Global Director, Business Strategy for Retail & CPG) and Atanu Ghosh (Director of Agentic Commerce & AI Innovation), who sit at the intersection of Azure AI, Copilot, and real-world retail and brand Agentic Commerce implementations.

This conversation dives into how Microsoft is thinking about agentic commerce as a full operating model shift, not just a feature or channel.
From Copilot shopping to data infrastructure to ads inside LLMs, the discussion connects the dots between:
Highlights 
  • Conversational discovery is fundamentally different from search: The way people talk about brands inside AI systems looks nothing like traditional keyword behavior, and most companies have zero visibility into it.
  • Agentic commerce is not one thing, it’s an operating model shift: Microsoft defines it broadly: agents participating in discovery, decision-making, and transactions, not just checkout.
  • Copilot + catalog + data = the new commerce stack
  • Ads are changing from persuasion to evaluation: Agents don’t care about creative. They evaluate structured data, trust signals, and relevance.
  • Loyalty becomes infrastructure, not a program
  • The “frontier firm” is the North Star: A future where humans and agents work together across workflows, not just AI acting as an assistant.
  • Protocols like UCP matter more than people realize: Open standards enable discoverability across surfaces and prevent fragmentation in agent-driven ecosystems.
  • The funnel is collapsing
If your brand is still thinking about traffic as the starting point, you’re already behind.
In an agentic world, discovery happens before you ever see the user, and increasingly, without them.

Timestamps
00:00 – Why brands have zero visibility into conversational discovery
01:50 – Scot’s “Big 6” podcast goal and Microsoft joins the list
04:00 – Michelle & Atanu backgrounds + roles at Microsoft
10:00 – What agentic commerce actually means (and why definitions vary)
14:00 – Microsoft’s agentic commerce stack: Copilot, agents, data layer
18:00 – The “Frontier Firm” and human + agent workflows
20:00 – UCP, protocols, and why openness matters
23:00 – Loyalty, identity, and friction in agent checkout
28:00 – Ads in agentic commerce: from persuasion to evaluation
33:00 – First-party data vs. conversational behavior data
39:00 – Why most retailers aren’t ready (and what they need first)
45:00 – The real blocker: operating models, not technology
46:30 – Where to learn more + how leaders are consuming information now

👉 Connect with Michele: https://www.linkedin.com/in/michelekfisher/
👉 Connect with Atanu: https://www.linkedin.com/in/buildingai/
👉 Learn more about Microsoft in AI: https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/the-future-of-ai-harnessing-ai-for-e-commerce---personalized-shopping-agents/4386980

🧠 Want to stay ahead in AI commerce? Subscribe and follow along:
📰 Subscribe to the free Substack: retailgentic.ai
📺 Watch episodes on YouTube & Subscribe for updates: https://www.youtube.com/@Retailgentic

Creators and Guests

Host
Scot Wingo
CEO and Co-founder, ReFiBuy.ai

What is Retailgentic | Consumer Behavior & Retail Trend?

Retailgentic, presented by the ReFiBuy Media Network and hosted by Scot Wingo, is where Agentic Commerce meets retail innovation. We help retailers and brands prepare for the future of agent-driven shopping with actionable news, expert insights, dxis, forward-looking predictions, and practical guidance on Agentic Commerce Optimization (ACO).

Speaker 1:

There is so little known about how people talk about your brand. But what we found out very early, even with

Speaker 2:

Copilot data, is that people were... The way that they were discovering brands through conversation was completely different and completely had upended how people used to use search. So adding a conversational component to a website doesn't have to just be like the AI agent that you're adding to your site so that you're with it. It's also capturing that conversational data and incorporating conversational data and how you're discovered into the entire process and the data estate.

Speaker 3:

Welcome to the Retailgentic podcast where we explore the intersection of retail, ecommerce, and AI agents, which we call Retailgentic. Our mission is simple. We are shining a bright light on the new Agentic AI shopping megatrend, which will disrupt the trillion dollar digital retail market. The stakes are high. Google, OpenAI, Microsoft, Perplexity, Amazon, Apple, Meta, and Anthropic are all fighting to be the consumer AI shopping agent winner.

Speaker 3:

There are gonna be some tough decisions for online brands and retailers to make. What happens to checkout? How should payments work? Could you continue to invest in retail media networks? What does the world look like when over half of your traffic is not from humans but from AI agents?

Speaker 3:

These existential questions are the tip of the iceberg. On this podcast, we're talking to industry experts, analysts, and influencers to get answers to these questions, identify trends, and best practices. This is Retailgentic, your podcast partner in navigating our AI shopping agent future. And now here's your host, Wingo.

Speaker 4:

Welcome to this episode of Retailgentic. When I started the podcast, I set a personal stretch goal to see if I could get the big six firms on. OpenAI, Anthropic, Perplexity, Microsoft, Google, and Meta in no particular order. If you're gonna do Agentic Commerce, you gotta have the big six. So far, we have been fortunate enough to have two off that list, OpenAI and Google.

Speaker 4:

Today, we're checking off a third, Microsoft. I'm excited to welcome Michele Fisher and Atanu Ghosh from Microsoft to the podcast. Michele is the global director of business strategy for Retail and Consumer Goods. Atanu is the director of Agentic Commerce and AI Innovation. Sounds like they're gonna know a thing or two about our world.

Speaker 4:

They are on a vertically focused team, that cuts across Microsoft, but it lives in the Azure organization or the cloud computing side of the house. In this conversation, we go over all the topics. So we we started the definition of agentic commerce. I've discovered that is important after my big debate at Shop Talk to make sure I'm not the only one that that thinks the way that the rest of the world does. All the various...

Speaker 4:

We also talk about the various Microsoft offerings that they have to offer across the different parts of the org. They have this interesting concept I thought was fascinating called a frontier firm that you'll hear about. We talk about the intersection of ads in addition to commerce, which is an area that I think is to be kind of a theme for 2027 largely. And then they're in hundreds of discussions every week with lots of really big brands, global brands, global retailers, and they're gonna share some of the the things they're hearing from all those conversations. I hope you enjoy this, talk, with the folks from Microsoft.

Speaker 4:

I learned a ton, and I just wanna say, anthropic, meta, and perplexity, you're next. Welcome to this episode of Retailgentic. I have not only one guest, but two guests, and they are joining me from this company called Microsoft. We have Michele Fisher. She is the Global Director of Business Strategy, Retail and Consumer Goods.

Speaker 4:

Welcome, Michele.

Speaker 2:

Thanks. Thanks for having me, Scot.

Speaker 4:

I always give my... I have another podcast with Jason Goldberg at Publicis, and he has a very long title. I think you may beat him by a word. We'll have to have a title off and figure out who has the longer title.

Speaker 2:

Oh, it's entirely possible. If you add the fact that I lead the strategy for marketing, CX and Agentic Commerce, then I think I would beat everybody.

Speaker 4:

We'll have to have you and Jason. We'll have to do something at a trade show sometime. Sounds good. And then we're also joined by Atanu Ghosh. He is the director of Agentic Commerce and AI Co Innovation.

Speaker 4:

Welcome. Welcome, Atanu.

Speaker 5:

Thanks. Thanks for having us, Scot. Good to be here with you, Michele.

Speaker 4:

Yeah. We have every time zone but mountain represented here. I'm East. Atanu is Central and Michele is Pacific. So next time we'll have to get a colleague from Denver and then we'll have it all covered.

Speaker 2:

Yeah, that's

Speaker 4:

true. One of my podcast goals I can check off my bucket list. Well, let's start. And Michele, we'll go ladies first. Let's start with your background, how you got here to Microsoft, and then help us understand your role at Microsoft.

Speaker 4:

I know you guys have a lot going on in the world and help us understand where you fit in.

Speaker 2:

Yeah, absolutely. So I've been at Microsoft for about six years now, and I spend my time with retail and consumer goods leaders trying to help them operationalize AI and specifically for CX marketing and Agentic Commerce use cases without just breaking their business. So that's a lot of my job and how I got here. I usually give just the standard. I've been working in retail for a really long time, couple of decades now in marketing and in ad tech.

Speaker 2:

And I actually was an opera singer before all of that. I think that is the most fun fact about me. So I I took the very natural progression into tech from there and, you know, traversed the the landscape of Disney and Amazon before I got here. So I'm super excited to have this conversation. I think this is the most exciting time that I've ever had in my career, especially from a marketer standpoint when I think about the entire question of who is our customer today.

Speaker 2:

We all want to tell the story to the right customer, and now the question is, is our customer human? So it's a fun time to be a part of this transformation.

Speaker 4:

Atanu, tell us about your background.

Speaker 5:

Sure. Happy to, Scot. Yeah. I definitely am not a singer. Definitely not an opera singer.

Speaker 2:

You have fun facts. You can deliver on fun facts.

Speaker 5:

I do. I think in threes. I think that's what it is. That's my secret part. I always think in threes.

Speaker 4:

Are you a triplet or something? Like, where where does that come from?

Speaker 5:

Yeah. So, yeah, that actually, I think, a muscle I've been. A lot lot... A large part of my career I've spent at Price Waterhouse Coopers, PwC, doing a of management consulting. So that's kinda where my career took off.

Speaker 5:

And as you can imagine, you know, working with one of the big four, you get to taste a lot of good experiences across a whole variety of customers. So, you know, finance, retail finance, travel, transportation, all of that. And I think what that that experience helped me shape for what I was seeing coming, which is the AI transformation, because a lot of my work was based in digital transformation, as we called it back in the day. And I said, hey. You know what?

Speaker 5:

I'd love to get on the front seat, the front row of this AI transformation journey. And that's what brought me to Microsoft. About the same time, Michele joined about six years ago. A lot of my work at Microsoft has been focused on the retail and consumer goods industry, largely helping customers do three things, and that's shaping hard problems to make them solvable, to help bring the best of our partners and platforms that we have to help them solve that problem right there. And thirdly, it's about strategic partnerships.

Speaker 5:

I think the way to succeed in this fast evolving space is to what we call derisk your burden of innovation with your partners and have them co invest with you. And that's a large part of the work that I do here. So I drive a lot of co innovation. I drive a lot of co AI transformation work for our customers, and that could be through setting up AI co innovation labs, co innovation hubs of that sort. So...

Speaker 5:

And lately, Agentic Commerce is what I've been diving deep into. So it's just phenomenal to see the resurgence of what we knew was e commerce, and it's been the same static flow for twenty plus years, and a new tectonic wave called Agentic comes in. So I'm super thrilled to be working in this space with our customers, and it's really exciting.

Speaker 4:

Microsoft has, obviously, have like your applications group, and that's where Office three sixty five is. You've got your online Bing and that group, where I imagine part of Copilot lives. And then you have your Azure side of things, and that's where the OpenAI partnership lives and obviously cloud computing type things. Do you guys sit in one of those, or does the retail and consumer goods piece, is it kind of an overlay that goes across?

Speaker 2:

Atanu and I sit within industry, and there's a group that follows all of the big industries that are connecting the dots between what's happening with Azure AI solutions and how the industry moving those into a go to market that supports how our retailers, consumer goods brands, our specific industries can use those technologies. So that's where we focus, is you take the people who have been in the industry for twenty, twenty plus years and help in strategizing how we bring our AI solutions to market for that industry.

Speaker 5:

That's great, Michele. I think what I can also add to that is a part of our mission is to also help our, what we call, our field, which is the folks who support our customers and helping them be more empowered in having these conversations. So we complement a lot of our teams that way. So we have the technical specialists who just know the whole tech stack of the world, if you will. And then on our side, Michele and I and our teams help supplant that with a lot of the industry context and depth to really help our customers make the right decisions.

Speaker 4:

I'll start with a kind of a... Hopefully, as a softball, we'll see how it goes. How do you guys define Agentic Commerce? I mentioned this, Michele, you were at Shop Talk, and I was on this debate thing with there's a couple of people that are kind of anti Agentic Commerce, if you will, the naysayers I call them. Their whole argument comes back to their definition of Agentic Commerce.

Speaker 4:

So I've learned that I need to back up a little bit on these interviews and make sure that we, you know, where do we all fall on this spectrum of definitions? I don't know who wants to go first, so I'll give you the jump ball here.

Speaker 2:

I noticed the same thing when I was at Shop Talk, Scot. And also, even when I'm talking to any one of our retail or consumer goods brands, I have found that it is incredibly different what how everyone is defining Agentic Commerce. How Microsoft defines agentic commerce, our definition is really in moving to an operating model. So it's that shift in how commerce works. AI agents are now first class actors in Discovery.

Speaker 2:

So decision making, they're even in transactions. But when brands are talking to us, how they define Agentic Commerce depends entirely on what they're responsible for within the company. So if you think about it on the merchant side, it's how do I connect my PIM data to my content so that I can ensure that the most relevant information is out there, discoverable, machine readable for agents to be able to capture so that my brand stays relevant. If you're talking to a CMO or a CX leader, for example, you might be talking about brand agents and how to build conversational agents into the actual structure of the CX, or how to capture all of the demand of the influx of a myriad of agents that are now going to websites to crawl and do discovery on behalf of a customer. So there's lots of different views of it, but how Microsoft is thinking about it is really deliberate and not getting trapped in a narrow definition and more thinking about it as equaling an agent that checks out for you autonomously.

Speaker 5:

I love that, Michele. Scot, to answer your question, I'm 400% with you where when you start a conversation at Agentic Commerce, you always wanna lay the definition because it feels like a lot of times we are speaking to people across a pane of glass. And we see the lips moving, but you're talking fast, speaking past each other. So I love this grounding question that you've asked. In my opinion, I think that the the definition of Agentic Commerce actually varies with the person you're speaking with.

Speaker 5:

Most of the CMOs we talk to, they'll... They they come to us saying, hey. I get this Agentic Commerce thing is happening. That's great. And there's so many different definitions and ways you could bring this.

Speaker 5:

But end of the day, I really care about personalization, acquisition, and conversion. How can you help me do that? And if Agentic Commerce is the solution to help them with that need, then that is the definition of Agentic Commerce. To me, really, it's about an enabling platform. It's a platform mindset that everybody has to pivot towards where the consumer can be met with their needs or what they desire to achieve at that moment, and the brand and the retailer should meet them on that platform to suffice that And that is the definition of what Agentic Commerce is.

Speaker 5:

It's consumer behavior. It's just a new consumer behavior. And we put technology behind it to help the consumer win.

Speaker 2:

And I'll add one. Also, is it a channel? Is it just a new channel that marketers, for example, have to think about when they're divvying up their buckets of spend with their agencies. So they're they're... Yeah.

Speaker 2:

It's it's a tough one. And I think the more broad you stay, probably the better you are off to fighting agentic Commerce.

Speaker 4:

That's a good jumping off point because I went to the Microsoft booth at Shop Talk and the amount... Number one, it's just like so impressive, like across our industry, how everything's Agentic Commerce now. But you go into the Microsoft booth and there's like... There's probably like 10 different ways you can interact with Microsoft at this layer of Agentic Commerce. There's, there's obviously like the Copilot, which is the consumer kind of, I think that's what you're calling the channel, Michele.

Speaker 4:

But then you guys offer a lot of, you could have an on-site agent, and you showed a lot there. You showed the ability to use LMs as kind of a, and I don't know what you call this, but there was like a Microsoft powered search engine type thing, you know, so a retailer could have a better search experience powered by Gen AI.

Speaker 2:

Yeah, grounding.

Speaker 4:

Yeah, on and on and on. So, yeah, pretty amazing. What... Maybe walk us through Microsoft's offerings and, you know, that come up when you're in these conversations with retailers and CPG companies.

Speaker 5:

So I think there's three offerings. And Michele, please keep me honest because I know we have many more that we also offer as part of our other parts of Microsoft, like Microsoft Search and Ads. Primarily, are three. The first one is obviously Copilot shopping, like you talked about, and there's a lot of enablers as part of that. You have the catalog enrichment that we provide, the grounding, which is a key new feature that we are helping all our customers get up to speed with the latest data that is out there because LLMs, again, as we know, have a Fresh Buy date on the the knowledge repo that they work with.

Speaker 5:

This new feature allows them to get the most relevant, fastest, zero latency, recent information. They need to enrich that insight. Then we have, for the brands, we have retailers, we have what we call brand assistants. And these are really conversational agents, conversational capabilities that work off the SendRich catalog to be able to help our brands surface the right product. And it's actually a way more richer experience because that's tied into the loyalty data.

Speaker 5:

It's way more personalized. It allows them to have that really interconnected experience and personalized experience with their customers and consumers. And then lastly, we have our... The ad side of it, which we have all the offer highlights and our, you know, ad optimization that we can do for our customers. But beneath all of that, I think what powers this entire stack is our data platform.

Speaker 5:

Because all of these systems, the the monolithic systems everyone lives and breathes by the PIM, the OMS, the DAM, they sit on a data stack, and we help agentify that stack so that everything above that layer works flawlessly, and it provides that experience. So I think those are in a box, if you will, the offerings that from Microsoft that customers can usually, you know, subscribe and use.

Speaker 4:

Does that data layer have a brand name or just the data layer?

Speaker 5:

It's a whole bunch of platforms. We have, you know, without getting into the product suite, we have Fabric, we have IQ technology, but these are really agentic services that allow data to be accessed, frankly. So we're moving away from the world of moving data from one container to another. It's about, hey, leave your data where it is. We'll find ways to access it.

Speaker 5:

So you're spending more time with the richer insights than just looking at data. So that's a whole bunch of products and platforms that support that. Did I miss anything, Michele?

Speaker 2:

No, just making the data available and what you do find available through a conversational interface and enabling Teams with that is a huge chunk of what we're talking about today. A lot of the questions that we're getting around everything from Agentic Commerce all the way to building out multi agent orchestration is around how do I change the teams to think about Agentic Commerce this way? How do I change the remit of the agent versus the human and how they're reacting to one another? So we have something called the frontier firm that we talk about a lot at Microsoft, where agents and humans are working together as colleagues and not just as AI assistants to get work done and to take on a lot of those menial tasks that will allow humans to do higher value work. So we have a myriad of different solutions around that, a lot of which is within our Copilot side of things.

Speaker 2:

And then there's also partners. We, on the side of industry, work a lot with our partners in figuring out how to connect the dots between the amazing solutions that they are bringing to the table with AI and also how to connect it into data state, connect it into Copilot, where you can co mingle it with the rest of your portfolio. So there's a lot of stuff at play right now.

Speaker 4:

So the frontier firm is kind of like a holy grail kind of aspirational kind of mythical, probably not the right word. Prototypical is probably the right word. So you use that as kind of like, you know, here's, you know, here is the, you know, the future state that, you know, in the frontier firm where you have humans with these roles and then they're supported and work with agents that have these roles. Is that kind of how you frame it?

Speaker 5:

Yes. And you're close. Yes. I think it's more a north star, the way we put it. To be frontier is more north star, and every marketing organization has about five, six main workflows.

Speaker 5:

And not every workflow is going to hit the frontier state at the same time. Some are more advanced, sophisticated, and more AI ready than others are. The journey to get there and close the gap is, I believe, what our focus is right now to help our customers with.

Speaker 4:

Is this something you guys have published? Can, yeah. Now I want to read about the Frontier Firm.

Speaker 2:

We've got a white paper.

Speaker 4:

Oh, white paper. I love good white paper.

Speaker 2:

Yeah.

Speaker 4:

So we'll, we'll put this in the show notes, and, we can all nerd out on it and learn about the Frontier Firm. That'd be great. Yeah. We're protocol nerds here on the podcast, and I know that's not 100% your value, but we'll talk about it big picture. It was really interesting to see Microsoft announce support for UCP, and there was a broader UCP announcement where they've added a lot of firms.

Speaker 4:

The most shocking was Amazon, and I know you guys can't talk about it, but everyone like my phone lit up totally when that happened. People got, you know, people were slightly surprised by you guys, and then they were like shocked by Amazon joining. So interesting set of things going on in our world. You know, to the extent you can talk about it, what's going on with you guys kind of launched with ACP, now you've added UCP. What's kind of the mindset there?

Speaker 5:

Scot, I think this goes back to Microsoft's core principle, I guess, ethos, that we are an open standard company. We are a company of open standards, and we want to make sure that we help our customers with the right sort of technologies that helps them do their business much easier than before. And part of that is adopting these sort of industry standards because we know customers would want to invest in that. It does not bode well as partners, strategic partners to a lot of our customers to have a proprietary integration layer, if you will, in this case, a protocol layer. That just means another integration point, more resources, more time, money, effort spent in maintaining that.

Speaker 5:

So yes, I think architecturally speaking, the commerce protocol, the UCP, I think, is definitely a well suited guardian, if you will, of what's coming in the Agentic world. We, in fact, advise our customers to be open to protocols and the commerce partners we work with, because these are very foundational investments that we know the customers can make today that allows them the flexibility and freedom more than anything else to be able to operate in a multitude of surfaces. Today, we know about a handful of three, four, five surfaces, and I'm sure there's going be more that come in and there's going be a world that all of that fold in. There's so much movement, but to have that foundation laid out allows them that speed and access to markets for the consumers that allow them to provide the right experiences that they need to get to the customers. So from that standpoint, yeah, Copilot is now from the Merchant Center, and you can turn on the UCP specs, and that allows more discoverability.

Speaker 5:

The customer still stays to be the merchant of record. That's usually an area of concern. But all of that is now being alleviated because of this new protocol and how we adopted that within our surface at Copilot.

Speaker 2:

I think it's also a key to adoption. When we look at how many people are actually transacting through LLMs and that number, it remains obviously much larger than it was a year ago. But it's certainly still there is a barrier to entry, obviously, when it comes to adopting transactions through an LLM across all industries and all platforms. So I think that not only do we welcome standardization, it reduces technical complexity for Microsoft and for our merchants as well. But retailers and consumer goods brands need to prioritize where to be wherever their customer is.

Speaker 2:

And agents are going to prioritize based on where the best answer, the best price is going to be for those customers. So there's a huge benefit to being discoverable everywhere that people shop. And Microsoft wants to help them retain that control. And same thing for the brands as well, retaining control of data, retaining control of intelligence that it generates in the customer relationships. So that's all very important to how we're thinking about this next phase.

Speaker 4:

What was interesting about the announcement was whenever I see these things, I immediately go and try it. And it was live that second, so that was interesting. So sometimes there's a bit of a gap between these things being announced and when they go live. So that was nice and refreshing to see it actually live. And then one of your launch partners was Target.

Speaker 4:

I was surprised to see just after the UCP recently, like basically in under a month, UCP announced some extensions to add a loyalty component through this piece called Identity. And you guys launched that with Target, where you can connect to your Circle account, which is we're all big Target people around here in North Carolina. So that was and Walmart. We love them both. And I thought that was really, really interesting.

Speaker 4:

I know you can't talk specifically about the Target relationship, but in a broader sense, are you hearing a lot? Was this a reaction to you guys hearing about loyalty and the importance of tying that together from the brands and retailers you deal with?

Speaker 2:

I I think that the way that we're seeing that is very similar, Scot. So the bigger story isn't any single retailer. Of course, we can't share anything that's not public, But it's really how loyalty and promotions are going to evolve with AI agents, how they're sitting inside that buying decision. So I think that what we're seeing here is essentially looking at what Agentic e commerce enables. That's the interesting, more broad shift.

Speaker 2:

Promotions become more of a signal. Promotions stop being static or campaigns, and they start behaving like real time decision inputs. And what I think we'll see also when it comes to ads especially is that agents aren't going to be swayed by the content that is available on an ad. That's going to be an evaluation of whether or not the product is right for the customer. So there's a lot of implications there around how that will impact loyalty, how to drive LTV, how to connect in the moment of intent and not after the fact.

Speaker 2:

And that unlocks a lot of what Agentic Commerce is capable of in loyalty and offers and ads that are layered on top of that customer journey. They're all part of how the agent is going to reason on behalf of the customer and act.

Speaker 5:

Yeah, I love that your point on what we call embedding loyalty deep in this experience, right? And we are in such early days right now that the experience that we all went through, I tried that myself to try and pick up a Barbie doll, if you will, at Target for my seven year old. Just like we had back in the day from a user experience standpoint, there is the login screen. You gotta tie in your information. And, yes, that is a much bigger reward for me as a consumer, a shopper at the end of it.

Speaker 5:

Yet, I find myself seeing that, okay, this is no different than us having the checkout box where you have to log in and punch in, like, 16 fields of information. For me, I believe the opportunity is if we can get away with that bit of friction because you have that high intent customer that showed up. They've clicked the buy button. They're so close to it. So how can you design these new systems between off-site commerce engines to retailers and brands where that goes away and it's just as seamless.

Speaker 5:

Is it through like setting up wallets, permission credentials for certain brands? And that then starts raising the question of, okay, how strong is your loyalty as a brand that I, as a shopper, will add you to my preferences within my off-site commerce wallet that, yes, if I'm shopping at Target, links it through. Walmart links it through. I think that's the new bastion for winning that level of loyalty and mindshare with a shopper. So there's no more clicks to start tying in rewards and loyalty and whatnot.

Speaker 2:

You bring up such a good point. I want to add on to that. So people don't think in work versus personal anymore. And I don't think that that's how they're going to want to act With Agentic Commerce in particular, they don't want to think about which agent they logged into to have that conversation. And so for brands in particular, that context across memory and across email and calendar, the shopping, the gaming, especially as more and more of us are working from home in many ways.

Speaker 2:

You've got people who are taking breaks and doing gaming, they're seeing a CTV ad and then typing in a question into an LLM and determining what their travel should look like in the next three to six months, and then also checking their email and prioritizing and optimizing their email inbox. So the great experiences that we're going to see forthcoming with this are going to be the ones that create that singular conversation that customers are able to have that includes memory, that includes the context of who they are, who their family is, how they work, what they're doing, how they game, essentially. And ID is going to be a huge aspect to that as well, keeping that consistent so that the customer experience remains consistent.

Speaker 4:

Yeah. It's a super nice feature of UCP that they've gone ahead and done some pretty good investments on that. So also, in addition to the loyalty piece, when you guys rolled out your UCP announcement, you announced a new ad format. It's called Offer Highlights. And I thought, as a catalog nerd, I thought that was interesting itself.

Speaker 4:

Always exciting to highlight part of the catalog or something unique about a product. In this example, it was Best Buy, and it was like, I think, a location based one. The... But I thought it was also... The bigger the the bigger story is, alright, so we obviously know OpenAI is gonna be investing in this, and then so is, Gemini.

Speaker 4:

But to see Copilot kind of take the the lead, even like Rufus has, sponsored prompts. So to see you guys kind of have a little signal saying, hey, we're gonna start exploring this intersection of of ads, and AI and Agentic Commerce. Talk a little bit broadly, like where do you guys, are you hearing that from brands and retailers? They want that? Or like, what are some of the things that you think are going on at that intersection?

Speaker 5:

Yeah, I think the intersection of ads and Agentic Commerce is a very interesting one and one that I wanna have very close eyes on. The offer highlights feature you're talking about actually really transformational in some ways because what it does is it takes what we know are the product differentiators that were loaded by the retailer in the Merchant Center and Copilot, and it surfaces that in context of the conversation. So it's not just a static, hey, this is what I'm looking for. If the customer is looking for these five things, these attributes go up. So it's actually woven into the context of the conversation.

Speaker 5:

That is a key differentiator. And I think it's that purpose built design, if you will, of the Soft Reliance that makes it more relevant as more retailers start to adopt that or brands start adopting that feature. But just stepping back in ads and Agentic Commerce, the world we humans know and we lived in is about the three things that we saw brands talk to us. It was attention. It was, you know, was attention.

Speaker 5:

It was persuasion, emotion and persuasion. Like, those are the three gates, if you will. In three years from now, even if agents were looking at Super Bowl ads, I don't know if they'd have a differing opinion saying, hey, this one is better than the other. For them, it comes down to just blobs and blobs of just highly enriched data, text, machine readable text. And I think what's gonna happen now is that tension that we had is gonna start migrating more towards all these brands investing more to make that more trusted, coming from a more authoritative source, and maybe even finding ways to have that sponsored so that that ad stream sits within the pipeline of selection for these agents.

Speaker 5:

I think that's going to be the new way in which ads and Agentic Commerce is going to blend in. I think we have moved from the model of the eras, if you will. We started Discovery, 16 clicks, 14 sorts, and 17 items later we check out. Then we moved into dialogue, which is, hey. Can you help me find something?

Speaker 5:

I think we're still at the cusp of figuring what delegation is. So it's a discovery, dialogue, and delegation. I think delegation is where the ads piece can add more momentum to help close and shrink that funnel going from awareness to purchase. I think that's really gonna be fundamental in seeing how ads push us that way.

Speaker 2:

I think ads are going to more and more stop being something that interrupt your journey and people in general as they're going through their Discovery experience. And they're going to be something that help people decide, especially with agents, not even people. So agents that are deciding on customer's behalf. So in Agentic Commerce, I think ads will be evaluated, not served. I think they'll show up as offers and availability.

Speaker 2:

I mean, this connection that we have now between the catalog and the way that people shop from an LLM perspective, means that there's such a wide variety of open opportunity for brands to be able to deliver competitive pricing or any type of loyalty through that dialogue. If that person is checking some boxes about how that they have shopped in the past that we haven't seen yet. People will be able to compare much easier, choose, act. And I think that we also have to acknowledge the fact that agents are already deciding how people shop with or without ads. My dad, for example, told me that he bought my Christmas present this year through ChatGPT.

Speaker 2:

And I didn't even know he knew what ChatGPT was. But it means that he had literally, I mean, I can't even describe what I do for a living to my dad. But what he had done is he has essentially put in some stuff that he knew that I liked and probably came up with three options. And those three options were so helpful to him and figured, you know, not having to do an actual search, but in the more traditional way. So that's the journey that ads are up against right now is taking that huge catalog.

Speaker 2:

And now you can already assume that an agent has helped make that decision and narrowed it down to two or three things. So that discoverability outside of that particular journey is where it's going to really be interesting for how ads evolve.

Speaker 5:

Yeah. And I love what you said, Michele. If you think about it, right, we speak to our customers every day and we only hear about the data challenges out there. Like there was already this huge chasm of first party data. Retailers, I think, are obviously a much better position because of loyalty programs, cards, this, that.

Speaker 5:

So you get that kind of consumer insight. Brands, it's always harder for them to get that first party unless they're enriching it with some third party data source. But if you think about it, now there's yet another new entire area of information that's coming up that contains this rich consumer behavioral data of how consumers are interacting on these LLM surfaces. And a lot of that is actually, if you think about it, first party consent in many ways. Like my LLM surfaces know what I shop for, who I shop typically with, and when I shop, what are the price points.

Speaker 5:

All of those markers are something that would be called preferential encoding. That's happened in these systems, and that's allowing them to select and choose. Brands need to start investing in that mindshare for the consumer so I can start adding them into that list of brands like this or items like that. And it's going to come down to three other factors that I believe customers need to focus on in the world of ads in Agentic Commerce, which is going to be trust signals, more publisher networks so their citations get optimized because that's going be a huge weightage that comes in moving from a rank. Now we're at a two decimal number in some parameter somewhere in the model.

Speaker 5:

And then lastly, it's the authenticity of that. So I think those forces have to be accounted for, and customers need to think about allocating budget to start investing in those spaces so that the ads that are served up feel relevant to that shopper, and they are aligned to what that preferential encoding was when the customers started shopping on that LLM with that agent.

Speaker 2:

Yeah, going back to your original question, Scot, about what is Agentic Commerce? I mean, even thinking about it from a retail perspective versus a consumer goods perspective, where you're thinking about it more from a monetization standpoint, for example, maybe you have retail media. How are you thinking about how that's showing up in your agent and how is your agent showing up in the LLM? But from a consumer goods standpoint, it's Agentic Commerce in terms of how I show up through a retailer agent and how I show up through an LLM. And both of those strategies are absolutely critical.

Speaker 4:

Yeah, you got to kind of optimize everywhere and it's Like, some of it's the same. Like, you know, the more attributes, the better everywhere. But the the, you know, the different context matters and different use cases, be it on a retailer's agent or one on your site or Copilot.

Speaker 1:

There is so little known about how people talk about your brand.

Speaker 2:

But what we found out very early, just even with Copilot data, is that people were... The way that they were discovering brands through conversation was completely different and completely had upended how people used to use search. Adding a conversational component to a website doesn't have to just be like the AI agent that you're adding to your site so that you're with it. It's also capturing that conversational data and incorporating conversational data and how you're discovered into the entire process and the data estate Atanu, that you were mentioning as well. How is that conversational data going to be leveraged for discoverability across the organization and customer insights, etcetera?

Speaker 5:

And Michele, I know this company called Refi Buy, Scot Wingo, that does some very good work.

Speaker 4:

Yeah, this is all we think about all the time.

Speaker 5:

Exactly. So yeah, I think it's about partnering with that sort of a capability, right, to be able to enrich your catalog with the right set of partners to help you understand how you optimize in the LLMs. I think that's key. And that feeds into the ad channel, the pipeline, and hence the funnel. Yes.

Speaker 4:

Yeah. And you got to get the context back from the destination whenever you can. Like, obviously you're not going to from certain of them, like chat's not going to give you back the conversation, but you get some hints through conversion data and citations and visibility and that type of stuff.

Speaker 2:

It's consumer insights and competitor insights is actually one of the number one use cases across retail and consumer goods in terms of where do I start my investment with AI? And how do we combine everything from world data, first party data, third party data, my demand sensing systems, etcetera, so that we can be forward thinking or even predictive about how we want to message to our customers. And that plays a lot into loyalty, what pricing should look at for customers. It's exciting to see how all of these different data streams are coming together into unified data lake or whatever it may be so that you can action on them and get insights even if you're nontechnical across the organization.

Speaker 4:

Yeah, that's a good segue into my next question. You guys, being at such a big company where you sit at the center of all this consumer and enterprise pieces and then also how people work every day, What are some other things you're hearing? And in this vertical where I imagine it's pretty easy for someone to pick up the phone and ask you guys for advice on these things, you're probably having five, 10 conversations a day, if not a week and, you know, hundreds between the two of you, you probably have a lot to triangulate on. So we've talked about loyalty. We've talked about some of the other, you know, data enrichment.

Speaker 4:

What are some of the other things that are top of mind out there with the retail and CPG folks you interact with on a regular basis?

Speaker 5:

Yeah. And Scot, you started the conversation saying, you know, there's some folks with contrarian points of views. While this is not contrarian, I think there's definitely a new dimension to consider. A lot of the customers we speak with, when we say, hey, let's talk about Agentic Commerce, it typically ends up becoming a conversation of, hey, I am running my e commerce stack with stuff I haven't updated endpoints on in the last seven years. I'm not ready for this yet to have agents come and shop on my side.

Speaker 5:

I'm not ready to have this personalized shopping. What I really need from you, Microsoft, is can you give me more Agentic in commerce? So that changes the framing. That has helped us understand how we better serve our customers because we are really helping them break down that layer and move away from all the hype and noise of getting ready for this new wave and really focus on the fundamentals, which is how can we help you build a Discovery Agent? How can we help you build a Catalog Agent?

Speaker 5:

How can we help you build a Publisher Network Agent? An Availability Promise Agent? So we're really helping them build

Speaker 2:

a And how do we get them to all talk to each other?

Speaker 5:

Exactly, yes. It helps people out there so that you you have platforms out there that allow you to do that. Ours does as well. And how do you use that technology so the customer is able to be more ready for that consumer that's gonna come and you can allow that? That I think is resonating way better with this Agentic in commerce perspective than like, hey, how can I just get my PDP all squared away, sit back, relax, and hoping that at a coin toss, I get selected?

Speaker 5:

But there's a lot more foundational work we're realizing that needs to happen to, like Michele was saying, right, like, the assets sitting in the DAM, you don't even know how well they're performing. And there's just more and more and more generative content being placed in the DAM. It gets pulled in the PIM. Was it the right one that got pulled in? So there's personalization that gets lost in that entire interconnectivity.

Speaker 5:

And customers are receiving that positively. They're like, yeah, that's exactly what I need right now. If you can add that Agentic capability for my teams, for my existing stack, then we get better ready for this new Agentic consumer or the agents that are gonna show them. So that's sort of how... That's been a broad thematic point of conversation with all my customers.

Speaker 5:

Michele, you have more as well.

Speaker 2:

Yeah, absolutely. I think there's been a real shift since about NRF, Scot, in thinking about AI investments as just driving productivity versus driving the bottom line and how to look at success metrics when it comes to driving revenue. So where I'm seeing a lot of the friction, and I agree with everything that you said, Atanu, but where I'm seeing a lot of the friction happening is peeling back the layers. Yes, there's the data issue. That is 100% you don't want to put AI on data that is not clean, accurate.

Speaker 2:

But also the change management that that will take. So when you were just talking through all of those different agents, the product agent, the content agent, a media planning agent, an insights agent, etcetera. Instead of it being a feature, what you want is a fully working agentic system. And in order to do that, it's going to take leaders at the top that are having conversations with each other and deciding what the center of excellence looks like for AI, security, governance, etcetera. So there's even in peeling back the outcomes that you want to experience as a brand, like I want to drive X percent higher revenue, you have to peel that back and actually start at the top of what is it that you're trying to achieve CTO?

Speaker 2:

What is it that I'm trying to achieve CMO, CIO? And how do we start merging what we're both trying to achieve with AI? And then how does that get translated back down to the VPs, to the managers, to the managers of managers, for example, and all the way down to the ICs. It has to be a top down initiative is what we've discovered at Microsoft. Even with our own transformation that we've been doing for the last three years.

Speaker 2:

So a lot of the questions that I'm getting from leaders in retail and consumer goods brands is also, how do we do this? We all know that in retail, love a legacy system. The thing's not broken. But there's a lot of questions. We're buying into AI as the pillar of transformation.

Speaker 2:

How exactly do we do that in a way that's effective for our teams and not only makes it efficient, but also drives bottom line?

Speaker 5:

We should add to what you said. Remind me of a thought. For a data platform company, there's a narrative that I've been using at least, I'll say, which is where we are debunking this adage that your AI is only as good as your data. And that coming from someone at Microsoft, people are like, wait, what did you just say? It's like your data, data.

Speaker 5:

And the reason for that is Agentic Commerce, at the end of the day, is a workflow. And really, what happens is that data follows the people that create and consume it. People follow a process that they're meant to work with. And in working that process, it creates more data. So there's this virtuous cycle that everyone is stuck in that hard to get out of.

Speaker 5:

Now we can work on the data, and we can change it. Yes, that's a whole new work stream. And we are working to change people, which the operating model that Michele talked about. But really, the weakest and yet the most powerful link in this trifecta is the process, the workflow. If you can redesign that, it unlocks everything else, and therefore it's actually the workflow that is more important than the actual data, because everybody has the data, and they know where the data is, they know what to do.

Speaker 5:

But you are asking customers to reframe that context to focus more on the workflow, and therefore Agentic Commerce fits in

Speaker 2:

Yeah, it's just like the question has shifted a little bit to how do I change my organization to not just think of AI as assistance, but what do I do when they actually start taking action autonomously? So that brings in the data piece. We actually have something called a work trend index that we publish it annually. And I believe there's twenty twenty six data available as well. But most companies aren't really blocked by the technology.

Speaker 2:

They're blocked by the operating models in which they're working. So 70% of leaders within that Work Trend Index and I think it's a giant, giant index, say that their company is experimenting with AI. So that's the other thing is operationalizing AI is a big challenge because only 30% say AI is meaningfully changing how work gets done. So when you think about that gap and then you compare it to how we're going to actually drive revenue with it, there's a change that needs to take place in order for one to impact the other.

Speaker 4:

We're up against time, and I appreciate you guys taking I know you're super busy, to get two of you is amazing. If folks want to learn more about what you all are up to, where can they go on the massive set of resources that Microsoft has? Is there like a centralized CPG and retail type area to go for? Or is there a centralized Agentic Commerce area people can go? Or just follow you on LinkedIn?

Speaker 4:

What do you recommend there?

Speaker 5:

I mean, a good starting point would be our Microsoft Search and Ads team, I believe. If you just look up Microsoft. If you look up Microsoft AI, you'll see a whole bunch of links that show up, especially for Agentic Commerce. There's a lot of publications. We have the work trend index that Michele talked about.

Speaker 5:

We have a white paper on search and discoverability. We have the white paper we just published you know, the whole Copilot offering that we have right now. Besides that, it would be LinkedIn, I guess. I know Michele and I, we publish and we share our thoughts pretty often, and we'd love to hear, and that's how we learn. So LinkedIn would be another place to find Michele and myself.

Speaker 2:

But definitely all the things you just said. However, I'm curious to know from both of you, where are you finding your news these days? Because I find it via an LLM, and I just cite the resources that I'm interested in. That's how I get my news these days.

Speaker 4:

Yeah. I have an agent that that knows what I consider priority sources and not priority sources, and it kind of sifts through all that stuff for me. At least gives me a prioritized list.

Speaker 2:

I see. No, I get so many emails from all of these people that I have followed over the years or different businesses that I followed. And so I'll just use a prompt that says go through all of the emails that I get into my view of your focus and your your extra type of emails and summarize to me what the biggest news headlines were and what did I miss, what's not in there that I should definitely know about.

Speaker 5:

It's definitely been an LLM. And I think, Scot, the work you published on Retailgentic is really good because of the recency. There is no latency in that information. So I think you did a really good job of just pushing updates out there really quick, so it helps us keep in the know. There's yourself, Jason.

Speaker 5:

We follow Jason quite a bit in the space as well. So, yeah, I think it's more of, like Michele said, like, just thought leaders, and I have an agent as well across our, you know, co work suite that we have that just scans and gives me the top five articles that were not discussed yesterday. So good starting point in the morning.

Speaker 4:

Awesome. Well, Michele and Atanu, really appreciate you guys coming on. And in the show notes, we'll put some references to all the sites you mentioned. And then also I'm excited to learn about the Frontier firm and see what those folks, those mythical folks are up to out on the frontier.

Speaker 2:

Awesome. Thank you so much for having us, Scot. This is... This has been really fun. I appreciate it.

Speaker 5:

Yeah. Thanks for having us, Scot. Really good conversation. Thanks, Michele.

Speaker 3:

Follow, rate, and share to stay up to date with Retailgentic, your guide to the future of AI shopping agents.