Episode Summary
Anna Luo, Chief Marketing & Customer Innovation Officer at DaVinci Commerce, introduces AI Commerce Innovators, the new AI Realized special series on agentic commerce that she hosts, and explains why she created it. She says personal AI agents can get to know a shopper over time and, once given permission, act on that shopper’s behalf, and she calls agentic commerce a fundamental change in commerce itself. She explains why the series starts with discoverability: where the digital shelf showed hundreds of products, an AI agent recommends two or three, and context is what gets a product there. She says agentic recommendations go down to the product SKU level, describes a way to bring product reviews and social conversations into a persistent knowledge graph, says the goal is humans in the critical loops rather than every loop, and lists the questions she hopes brands will be asking a year from now, including how sales lift is measured.
Key takeaways
Ask what a personal agent is allowed to do and who is responsible when something goes wrong. Anna Luo says the questions become privacy, authority, security, alignment and accountability, including whether an agent is really acting in the shopper’s interest, and that the series will take these up with experts
Plan for agentic commerce as a change in commerce itself. She says AI can interpret intent, make recommendations and increasingly take actions on people’s behalf, so the mistake would be to treat it as simply another channel
Put humans in the critical loops rather than every loop. She says humans in every loop would defeat much of the value of the automation, and that the goal is humans in the loops that could change the course of judgment, ethics, brand decisions, risk or accountability
Aim for your product being recommended, beyond your brand being mentioned. She says that before an agent can buy for a shopper, a product first has to be discovered, and that in agentic commerce what you want to see is your product being recommended
Compete to be one of the two or three products an AI agent recommends. She says the shift is from hundreds of products shown on the digital shelf to two or three recommended, and that context is what gets a product there
Expect AI agents to recommend at the SKU level, beyond the category. She says recommendations go down to the product SKU level, and that a product detail page lists attributes that are fairly narrow and would not accommodate every question someone could ask, such as whether a foundation looks orange and cakey after a hot day
Where you can, look beyond the product description to signals such as reviews and social conversations. She describes ingesting how people use a product into a context memory, a persistent knowledge graph that can serve the response to the exact question asked and surface the product
Move from asking about AI strategy to asking how discoverability is attributed. She hopes brands will ask whether their products are discoverable, how to move from discovery to recommendation to sales lift, and how that sales lift is measured
Experiment now rather than waiting for everything to become obvious. Among the questions she hopes brands ask is how to experiment now, because by the time everything is obvious it might be too late
About Anna Luo
Anna Luo is Chief Marketing & Customer Innovation Officer at DaVinci Commerce, formerly Jivox, and the host of AI Commerce Innovators, the AI Realized special series on agentic commerce, whose first episode is AI Commerce Innovators Ep. 1: Nobody’s Browsing Anymore. She has spent more than two decades building new technology for enterprise customers, from real-time data platforms to one-to-one personalization at scale, and the last 10 years at DaVinci Commerce, where she leads an AI-powered growth team and works with brands to pilot agentic commerce. She is a member of the AI Realized leadership team.
In this episode
| 00:00 | Welcome and the AI Commerce Innovators series |
| 01:23 | From AI assistants to agents that act |
| 02:10 | What makes a personal AI agent different |
| 04:19 | Privacy, authority, security, alignment and accountability |
| 05:06 | The opportunity: from search to expressing intent |
| 06:39 | A fundamental change in commerce itself |
| 07:24 | Who listeners will hear from in the series |
| 08:39 | Bring your questions and join the conversation |
| 10:03 | Humans in the critical loops |
| 11:06 | Why the series starts with discoverability |
| 11:55 | From hundreds of products to two or three recommendations |
| 13:47 | Down to the product SKU level |
| 14:30 | Reviews, social conversations and a context memory |
| 16:10 | The questions brands should be asking a year from now |
| 17:00 | Where to automate, and why to experiment now |
| 17:41 | Where to follow AI Commerce Innovators |
| 18:55 | Curiosity with judgment |
| 19:14 | Stay accountable for the decisions |
In her words
“It isn’t adding AI to commerce. It is a fundamental change in commerce itself.”
Anna Luo
“AI can dramatically expand what people are capable of doing, but we shouldn’t outsource judgment to autonomous AI just because we can.”
Anna Luo
“Curiosity keeps us exploring, and I think judgment keeps us from blindly following technology.”
Anna Luo
“Use all the intelligence available to you, but stay accountable for the decisions.”
Anna Luo
Resources
The guest
Anna Luo on LinkedIn: Host of AI Commerce Innovators and Chief Marketing & Customer Innovation Officer at DaVinci Commerce
DaVinci Commerce: The company where Anna Luo is Chief Marketing & Customer Innovation Officer and works with brands to pilot agentic commerce, and the sponsor of AI Commerce Innovators
DaVinci Commerce on YouTube: Where AI Commerce Innovators episodes are published as video
Ideas and terms discussed
Agentic commerce: Commerce in which AI can interpret a shopper’s intent, make recommendations and increasingly take actions on the shopper’s behalf
Humans in the critical loops: Anna Luo’s goal for automation: humans in the loops that could change the course of judgment, ethics, brand decisions, risk or accountability, rather than in every loop
Context memory: A persistent knowledge graph built from product data, reviews and social conversations, which Anna Luo describes as serving the response to the exact question a shopper asks
Related AI Realized episodes and events
AI Commerce Innovators Ep. 1: Nobody’s Browsing Anymore: Anna Luo with leaders from Accenture, Forrester and DaVinci Commerce on how products get discovered and chosen when shoppers ask an AI.
AI Search Visibility: When AI Says Your Company Is Dead: Curtis Sparrer of Bospar on how AI answer engines decide what a company is, and what to do when one gets it wrong.
AI Turns Brand Valuation Into a Quarterly Management System: Hampton Bridwell of CorebrandAI on rebuilding brand valuation with AI, and why discoverability in AI platforms matters to brand value.
Your AI Agent Is Not the Risk. Its Authority Is.: Yogita Parulekar on identity and permissions for AI agents, and what an autonomous agent is permitted to do once governance exists.
Frequently Asked Questions
What is agentic commerce?
Agentic commerce is commerce in which AI can interpret a shopper’s intent, make recommendations and increasingly take actions on the shopper’s behalf. Anna Luo, Chief Marketing & Customer Innovation Officer at DaVinci Commerce and host of AI Commerce Innovators, says the mistake would be to think of it as simply another channel. In her words: “It isn’t adding AI to commerce. It is a fundamental change in commerce itself.”
How is a personal AI agent different from a shopping assistant?
A personal AI agent differs from a shopping assistant because it can get to know a person over time. Anna Luo of DaVinci Commerce says such an agent may know someone’s preferences, past purchases, accounts, loyalty programs, calendar and schedule, and potentially even health and financial information, depending on how much they are willing to share. Once given permission to act on someone’s behalf, she says, it is no longer just helping them find things.
What questions do personal AI shopping agents raise?
Personal AI shopping agents raise questions of privacy, authority, security, alignment and accountability, according to Anna Luo of DaVinci Commerce. She asks what the agent is actually allowed to do, whether its decisions can be manipulated, whether it is really acting in the shopper’s interest, and who is responsible when something goes wrong. She says the AI Commerce Innovators series will take these questions up with experts.
Should humans be in every loop when AI agents automate work?
Humans should be in the critical loops rather than every loop when AI agents automate work, according to Anna Luo of DaVinci Commerce, because humans in every loop would defeat much of the value of the automation. She says the goal is humans in the loops that could change the course of judgment, ethics, brand decisions, risk or accountability. In her words: “AI can dramatically expand what people are capable of doing, but we shouldn’t outsource judgment to autonomous AI just because we can.”
Why does product discoverability matter in agentic commerce?
Product discoverability matters in agentic commerce because before an AI agent can buy for a shopper, the product first has to be discovered, and what a brand wants to see is its product being recommended. Anna Luo of DaVinci Commerce says that in digital commerce the goal was visibility on the digital shelf, where hundreds of products are shown, while in agentic commerce an agent recommends two or three. Everybody is competing for those recommendations, she says, and context is what gets a product there.
How can brands get their products recommended by AI agents?
Brands can get their products recommended by AI agents by giving them product information, down to the SKU level, that goes beyond the attributes on a product detail page, according to Anna Luo of DaVinci Commerce. She says product detail page attributes are fairly narrow and would not accommodate every question a shopper could ask, such as whether a foundation will look orange and cakey after a hot day. Where there is a way to look at product reviews and social conversations to see how people use a product, she describes ingesting that into a context memory, a persistent knowledge graph that can serve the response to the exact question asked and surface the product.
What questions should brands ask about AI and commerce?
Brands should ask whether their products are discoverable and how that discoverability can be attributed, according to Anna Luo of DaVinci Commerce, who hopes brands move beyond asking “What is our AI strategy?” She hopes they will also ask how to make product data understandable and readable by personal AI agents and retail shopping agents, how to move from discovery to recommendation to sales lift, and how sales lift is measured. She adds the questions of where to automate, where humans should remain accountable, and how to experiment now rather than waiting for everything to become obvious, because by then it might be too late.
What is the AI Commerce Innovators podcast?
AI Commerce Innovators is a special series on the AI Realized Podcast about agentic commerce, hosted by Anna Luo of DaVinci Commerce. Its first episode featured Emily Pfeiffer of Forrester, Chip Knicker of Accenture and Diaz Nesamoney of DaVinci Commerce, the series sponsor. Anna says listeners will hear from practitioners, brands, retailers, direct-to-consumer companies, thought leaders, innovators and analysts, and she invites them to bring their questions and join the conversation.
-
[00:00] Christina Ellwood: Welcome to AI Realized, the podcast for enterprise executives leading AI adoption. From tackling security, data, and operational challenges to navigating organizational transformation, AI deployment offers a unique opportunity to redesign our organizations from the inside out. I’m Christina Ellwood, your host for today’s episode, and today is a little different. We’re introducing a new special series on AI Realized, the AI Commerce Innovators Circle. AI is changing how people discover, evaluate, and buy. More shoppers start by asking an AI assistant what to buy, and the assistant answers with a short list of recommendations. For brands and retailers, that raises an urgent question: Will your products be the answer? The host of the new series has spent more ti- more than two decades building new technology to enterprise customers from real-time data platforms to one-to-one personalization at scale. She spent the last 10 years at DaVinci Commerce, formerly Jivox, where she leads an AI-powered growth team and works with the gr- the brands to pilot agentic commerce. She’s gonna share why she created this series, who you’re gonna hear from, and what leaders will take away from it. Anna Luo is chief marketing and chief innovation officer at DaVinci Commerce. Anna, welcome to AI Realized.
[01:17] Anna Luo: Thank you, Christina
[01:18] Christina Ellwood: What inspired you to start this podcast series on AI commerce, Anna?
[01:23] Anna Luo: Where should we start? AI is getting better every day. Within a short period in commerce, it is evolving from generative AI to agentic. And it seems like only yesterday that we talked about an AI assistant like ChatGPT about our problems, our needs. We use it to research and get recommended solutions, and often it includes product recommendations. So that’s it. We, the humans, actually take actions to buy and decide where to buy. We still do that, and now we see OpenAI has Dot, right? And Meta has Muse, and there’s also Instinct and a whole series of other emerging AI personal agents. These are agents that take actions on our behalf. So I think what makes a personal AI agents different from shopping assistants agents that we’ve seen today is that it can actually get to know us over time. It may know our preferences what we’ve bought before, our accounts, loyalty programs, our calendar, our schedule. So potentially even things that are health and financial information, all that information. It depends on how much you’re willing to share with it. And once we give it permission to act on our behalf, it’s no longer about just helping us find things. It becomes an economic actor representing us, which you explain in the the article that you shared out. So this creates tremendous opportunity, but it also raises some pretty important questions. I see this podcast as a way for us all, for us practitioners to look deeper at how to use this very important tool to further human interest and progress and making commerce more efficient for shoppers, for brands, and for retailers. As we all know, AI Realized hosts roundtables for practitioners to share AI use cases, case studies agent controls, and governance gap. These are excellent topics to raise our awareness and learning, and in this podcast, we explore these questions through the lens of agentic commerce. For example, how do you balance hyper-personalization with privacy efficiency with control, and also automation versus human judgment?
[03:51] Christina Ellwood: I see. Okay. So the podcast is-- sounds like is going to f- talk both about the opportunity and the alignment of the c- technical capabilities to the goals of helping buyers buy. But it’s also gonna be featuring, it sounds like the discussion about what the operating system needs to look like in order to manage and control that new capability. Is that right?
[04:19] Anna Luo: Yeah. Here’s how I think about it. It on the one hand it can be very productive and efficient. But AI, as you have been hearing a lot of the recent reports, from Anthropic that AI sometimes can go rogue. And the questions become privacy, authority security, alignment, and accountability. Who controls our personal contacts? Who is the agent actually allow-- what is the agent actually allowed to do? Can its decisions be manipulated, for example? Is it really acting on our interest? And when something goes wrong, who is responsible? So these are some of the topics that we will very soon be talking to exort- experts about. And at the same time, the opportunity is just as big, right? Personal agents could completely change how people shop from search for products to simply expressing intent and agentic personalization to individualized recommendations. And eventually they may transact directly with the brands. So I think consumers have actually a lot of quite a bit of control over the direction because our behavior will shape, what the direction of these AI agents. So commerce is becoming more conversational, more personalized, and increasingly agentic. And today, because nobody has the playbook, so I want AI Commerce Innovators to be a place where people building this future can learn from one another.
[05:56] Christina Ellwood: That’s fantastic. And in your work in commerce, you have been involved in real-time recommendations and one-to-one personalization, and now agentic commerce at DaVinci Commerce. What’s that taught you about that someone that’s new to agentic commerce might miss?
[06:15] Anna Luo: So this is what I’ve learned. In every tectonic shift in technology that starts with the technology the real transformation actually happens when behavior change, as I just mentioned. So in commerce, we saw transformation with digital, mobile, and personalization. And AI is different because it doesn’t just change the interface. It can interpret intent, make recommendations, and increasingly take actions on our behalf. So the mistake would be to think this is simply another channel. And just to be clear, it isn’t adding AI to commerce. It is a fundamental change in commerce itself.
[07:00] Christina Ellwood: That is a very powerful differentiation that you’ve just just outlined for us from the past technologies. What kinds of leaders will listen-- or listeners hear from in this series? Who will you be featuring, and what should they expect to walk away with as things that they can apply or things that they can think about after the episode has ended?
[07:24] Anna Luo: So let me start with who we’ll be featuring. As we have already started with the Forrester analyst, Em-Emily Pfeiffer and also the Accenture agentic commerce lead Chip Knicker, and also our sponsor’s CEO, Diaz Nesamoney. So that would be the-- that was the first look into what is the state of the agentic commerce. And I think coming up, we will also be interviewing experts for example the lecturer of UC Berkeley engineering department. He has authored a lot of articles about chain of thought reasoning and misalignment. So some of those can address a lot of the questions about privacy and security. And we’ll also be talking to, LinkedIn top voice people who have been working in the, the various verticals of commerce and get their perspective about where they see the transformation is happening. So there’s gonna be a lot of interesting and very yeah, I would say expert lineup that could give us a, a very broad perspective of what’s happening in agentic commerce.
[08:32] Christina Ellwood: And so what should readers or listeners what should they expect to walk away with?
[08:39] Anna Luo: I would say come with an open mind, have a conversation, listen to what the experts are saying, and feel free to comment in what you’re thinking in response to what you’re hearing the experts are saying. It’s an open dialogue and because it’s so new, everything in AI is evolving almost on a daily basis. So we all can be considered experts if we’re all in commerce, agentic commerce. So I would say don’t think about just what you’re walking away with, but also bring your questions, bring your thoughts, join the conversation.
[09:18] Christina Ellwood: Great. Anna, you describe the AI Innovators Circle as being for leaders that are responsible for deciding how AI is adopted. What kinds of leaders will our listeners hear from in this series?
[09:32] Anna Luo: Yeah, Christina. So these will be experts like practitioners, brands, retailers, direct-to-consumer companies, also thought leaders, innovators, analysts. Basically, people who are moving the technology forward.
[09:43] Christina Ellwood: That makes a lot of sense. It is really a rapidly changing area. In fact you lead an AI-powered growth team at DaVinci Commerce, so you live every day this rapidly changing technology. As you’re balancing AI automation with human judgment, what does that balance look like today?
[10:03] Anna Luo: I don’t think the goal should be humans in every loop. That would defeat much of the value of the automation. So the hu- the goal is humans in the loops that could change the course of judgment, ethics, brand decisions, risk, or accountability. AI can dramatically expand what people are capable of doing, but we shouldn’t outsource judgment to autonomous AI just because we can. So I want to emphasize that the goal isn’t humans in every loop. It’s humans in the critical loops. How do we define critical loops? That could be another episode.
[10:40] Christina Ellwood: It could be another episode. I love that. In fact, much of the conversation has been about agentic com-- about agentic commerce has been about AI agents buying for us, buying on our behalf, if you will. But you chose to start this series not on that topic, but on discoverability, whether an AI recommends a brand’s product in the first place. Why did you start there?
[11:06] Anna Luo: Before an agent can buy for you, for-- you need to first be discovered. And so that’s why discoverability needs to happen. And the difference, as you may be seeing now, is you’re not just gonna be-- it’s not gonna be discover in the sense of AI visibility, that your brand gets mentioned. In agentic commerce, it’s about sales. More than just being mentioned, where you want to see happen is your product being recommended? And so let’s t- look at the shelves, right? For the longest time, we’re looking at shelves. So during the digital age, digital commerce it is about search and it is about your product showing in the sh- visibility of the digital shelves. Whereas in agentic commerce, what we’re seeing is the recommendation. So you’re all of a sudden moving from hundreds of products being shown to just being recommended two or three products being recommended. So everybody is competing for that two to three products being recommended. And what’s gonna get us there is the context. The users have so many different ways of asking questions in their prompts. They talk about their needs, their problems, how to, like what kind of solutions, and eventually is what kind of product can help me solve this problem. So from that perspective, the context matters hugely, and through that context is how the product is gonna surface
[12:36] Christina Ellwood: I see. So I guess I, I didn’t understand the idea that the shelf got smaller. So when we talk about the shelf, we’re talking about when I walk into a store, what do I see on the in the aisles, on the shelves and in the face fronts of the products and things like that. When you move that to digital, like e-commerce, the equivalent of that is the array of products that’s shown to me on the screen at one time. R- right? And now you’re saying with agentic commerce, I’m only gonna see the top three products that meet the criteria or whatever. So the shelf just got really small, but the shelf now is very specific. When I walk into the store and I look at a shelf of, whatever the product is, let’s say it’s cosmetics, I see lots and lots of mascaras and, lots and lots of every category of product, and I’m able to digest the shelf based on category. And in the case of e-commerce, the categories are more narrow, but in the case of agentic commerce, it’s basically down to whatever category meets my prompt. Is that a correct interpretation of what you just said?
[13:47] Anna Luo: It’s more than just categories. We’re actually going down to the product SKU level. There is-- So you may-- So in a, product dis- detail page, it might say this mascara has certain attributes, but these are fairly narrow and it wouldn’t accommodate all the possible questions somebody could ask. So for example, is this mas- foundations, right? Some of us who come to the show, we we put on makeup and so the foundation, it could be, some type of foundation might-- Some people might be allergic to the certain types of chemicals in the foundation and also certain foundation after a hot day, or being in the light for a long time, it might look orange and cakey. And so these are some of the examples that we use when we do the storytelling and to our customers. And so what we’re seeing is where do you find the, The right-- to h- where do you, how do you surface the right product when a certain prompt is asked? So beyond just describing, looking at the product description if there is a way to look at product reviews, for example social conversations, and see how people use these products. And so from there, you can ingest all of that into a, a, for example, products, a context memory where it’s a persistent knowledge graph where if it has certain, if any question is asked, you can just pull the memory could just be able to serve, the context memory be able to serve exactly the response to the question that was asked and surface the product.
[15:25] Christina Ellwood: As a consumer, I love that because I am not a shopper, I’m a buyer. And so you’re telling me that I’m gonna get exactly the product that I’m looking for and only have to choose between a few, and I can buy it immediately without ever having to leave my desk. That sounds like a nirvana to me. I’m very curious, Anna a year from now, what do you hope these leaders of agentic commerce are doing differently? Or I’m sorry, what are you hoping that these listeners who are deciding how AI is adopted in their organizations what are you hoping they’re doing differently because of the conversations that you’ve held with these experts?
[16:10] Anna Luo: One thing I hope for is that brands would move beyond asking, “What is our AI strategy?” Or “How are we being mentioned in, in, in the LLMs?” And my hope is that they would ask whether the products are discoverable as to what we just descri- discussed, and how discoverability can be attributed. So questions like, how should we think about using technologies to make product data understandable and readable by a personal AI agent or retail shopping agents? How does AI change not just how our customers discover us, but how to move from discovery to recommendation to sales lift? So at the end of the day, commerce is all about sales, and how is sales lift measured? And also where should we automate? Where should humans remain accountable? And how do we experiment now rather than waiting for everything to become obvious? Because at that point it might be too late.
[17:14] Christina Ellwood: Yeah. I’m actually really interested too in the relationship between consumers and brands, because this is a really different way to experience a brand. So those sound like really interesting things to listen for and look for in this wonderful new podcast series. Anna, what recommendations do you have for listeners, resources that they could use to learn more about you and about AI Innovators Circle?
[17:41] Anna Luo: I think the “AI Commerce Innovators” podcast will be a really good place to start because this is all top-of-mind issues that we all experience in agentic commerce, and we are looking for expert-- people, practitioners, like we talked about before, and different kinds of experts to come and share with us what they’re seeing from their-- where they, from where they sit. And so to subscribe to “AI Commerce Innovators,” you can do it on the “AI Realized” podcast or DaVinci Commerce YouTube channel. And we’re bringing together the people shaping this transition and making the conversation practical for brands direct-to-consumer merchants, and retailers.
[18:25] Christina Ellwood: One thing I should mention is that if you do subscribe to the YouTube channel, you’ll be experiencing the video of the podcast, whereas “AI Realized” is an audio podcast. So if you like to watch rather than listen only definitely subscribe to the DaVinci in DaVinci Commerce- Commerce ... AI Innovators Circle YouTube channel. Anna, in the AI revolution, what leadership skill do you find most valuable in your work today?
[18:55] Anna Luo: I would say curiosity with judgment. Curiosity keeps us exploring, and I think judgment keeps us from blindly following technology. So these are the two things.
[19:06] Christina Ellwood: Good answer. So if our listeners remember just one thing from today’s conversation, what should it be and why?
[19:14] Anna Luo: I would say don’t outsource your judgment to AI. Use all the intelligence available to you, but stay accountable for the decisions. So until next time, innovate responsibly.
[19:26] Christina Ellwood: There we go. Okay. Anna Luo, Chief Marketing and Customer Innovation Officer at DaVinci Commerce, and host of the AI Commerce Innovators Circle. Thank you so much for sharing your experience with us today.
[19:40] Anna Luo: Thank you, Christina.