The Agentic Commerce Stack — Ahnaf Prio, Best Buy

Figuring out what agentic commerce means at Best Buy

00:12My name is Anaf Prio. Uh I'm a senior engineering manager at Best Buy and me and my team are working together right now to figure out what does Agentic Commerce mean and how can we meet our customers where they're at and the newest place that they're at is at Agentic Services.

00:27I'm excited to give my talk today and well what's what's my credentials where ever since I was a young boy I dreamed of high throughput inference harnessing my tools within a context window kept in check with evals yeah that's absolutely correct in 2003 all those things definitely existed I kid uh over the last one year uh we have been learning a lot shopping isn't new shopping is probably one of the most fun things one can do and one of the most essential things that people need to do ever since uh the economy existed.

00:59But I have been super excited by it. So I'm going to go talk about what are the things some of the things that I've learned and hopefully uh share the notes. So what is agentic commerce? I'm not going to go over the broad definition again, but basically it's the idea that AI assists will help you with your shopping journey.

01:23Shopping has different facets to it. For instance, there's discovery. There's figuring out

Shopping is 45% of agent sessions

01:31the aspects of do I actually truly need it? Understanding and deciding, there's loyalty. There's pricing. There's fulfillment, post fulfillment. It's a lot. And believe it or not, right now about 45% of all agent sessions that happen within major providers like chatgbt.com and Google Gemini are related to shopping.

01:48Maybe it's a little biased uh that I don't use it as much, but I'm an engineer, but the humans out there are using AI and talking to them to help with their shopping journey. So it's also not a binary, you know, uh right now we're at that state of human in the loop.

02:12The ideal state would be autonomous shopping. You tell what you're excited about. Your agents goes around, talks to different merchants. Uh I'm originally from Bangladesh. We haggle a lot with the merchants too. Maybe does that negotiate, does the payments, but right now we're in the human in the loop.

02:27And the talk today is going to talk about the mental model of how that human in the loop is working right now while also provide you with the architecture if you choose to extend it or show you the vision of how autonomous shopping might work.

02:41So this isn't our first first attempt. Uh even a year ago there were people trying to figure out how can we automate this. Even now you can go probably download the cloud chrome extension. Maybe you've used Atlas where you tell the AI you need something.

03:00You need headphones. You need uh that that

Why screenshot and DOM agents failed

03:08grocery list of items that you have been meaning to buy but never made the actual effort to show up because you know you didn't have the time. So why don't you take screenshots, read the DOM, navigate to the merchant site, fill forms for me, do loyalty.

03:23It kind of just didn't work as expected. It was really clunky and slow and brittle. And if you are a merchant who's trying to sell stuff, any engineering department of that merchant will tell you an AI impersonating or your browser is just firing up all the alarm bells.

03:45So a lot of times you will probably be even stuck on the payment flow because we don't want you to be using AI to put in that order or at least in that phase that's what was happening. So what did uh what did it actually work and is it actually working right now?

04:00It is uh chat GPG shopping Google AI mode is doing just that. Uh right now agent shopping is considered to be a 7 billion industry and might go up to 65 billion industry by 2030. And the majority of the shoppers are using the mainstream conversational AI assistance which is on their browser or in your app chat GPT and Google Gemini.

04:28We're also seeing that pop up in Instagram and Facebook. Meta Meta wants to do meta commerce now. I heard go puff and uh Grock came together to make an app as well. And also Microsoft Copilot just yesterday announced in the UK that you can buy Ray-B bands now inside Microsoft Copilot.

04:49So to make that happen, Google and uh OpenAI separately came up with their own

Why merchants needed ACP and UCP

04:56little primitives ACP and UCP, which is basically talking about how you would actually talk to us. For some of you who are shopping on the other side as the customer, there is not much of a difference between adding an item to cart, adding a second quantity.

05:04But to us merchants, that's a second line item, buddy. That's not the same skew. So if we don't talk about the nuances and the primitives of commerce and standardize it, things will just not work and will remain to be clunky. So ACP was chat GPT's attempt at it and universal commerce protocol UCP was Google's attempt at it.

05:27So now that I've already established that this is happening, I just wanted to say that it is happening as easy as you go to the chat. GPT Gemini, tell it to find me cat cookies. more to it. Uh why I chose cat cookies later. Uh in this example, the AI surfaces the product.

05:45Agent calls the merchant checkout API. No browser payment flows via scope payment mandate or a delegated payment token. An order confirms and human kind of didn't have to touch the cart. So to all of this that's happening for the user, a lot is happening on the other side.

06:06And it's kind of overwhelming. One day we're talking

The acronym map: MCP, A2A, ACP, UCP, AP2

06:14about MCPS, another day A2A, ACP, UCP, AP2. Is like what is even real? Like if someone came up to me tomorrow and said, I came up with HYP, I would probably think it's probably real. So I wanted to dissect this mental model for you as I've learned about it more.

06:29MCP is still the model context protocol, the way that the AI agent identifies the tool. So maybe we can figure out what does this AI agent uh uh specifications are to showcase what products they have to showcase the details of a specific product to showcase loyalty.

06:46A2A is how agents talk to each other. They're more of a spec. ACP UCP are the primitives and AP2 is the agentic payment protocol payment mandate that Google's open specification came out and we'll talk all of them one by one how they actually relate to shopping.

07:02So the MCP tool access is very important because without knowing the different capabilities and hitting those different capabilities, taking the time to bring it into context, understanding the user's memory, the the agent will never be able to figure out what you're even trying to do.

07:19And the only way to get access to the specific capabilities is through MCP tool calls. Uh the next one is A2A. So now there are different ways to architect this different uh capabilities I talked about like payments uh let's say uh what do you call it loyalty you can make agents about specific domains itself sorry uh uh right and if you have specific domain level agents agents need to talk to each other we need to find a standardized way to talk to each other so A2A those specifications kind of fill in that gap uh also if you're a customer

08:00agent and your merchant agent uh need to talk to each other. Maybe you could they're both agents maybe we can use A2A. So now to the UCP MCP primitives. So the most important data is that product data. So UCP allows for adding that product data in a more uh more organized

Product feeds instead of catalog search

08:21way and ACP does the same because we don't want to go through your PDP and crawl and figure out every specific attribute. merchant just tell us and also those pro products change a lot so maybe you can tell us when they change as well to send this so that kind of data is happening uh uh that that kind of data flow is happening in the product feed normally you would assume that this would be a search catalog however both ACP and UCP right now so Gemini and chat GPT does not support that search catalog call they want you to send that feed to them and for those of you who are like

08:57why wouldn't you do that there's reasons to it sponsor products retail media of related things ranking but the most important technological challenge is if you have m number of merchants and n number of products now it has to call that many while if you send the product feed ahead of time we can index it and be ready offload to offload when you ask for something that I've also put an example of meta's product feed as you can see they're similar but still different everyone has an opinion they think their opinion is the best one and That's what they're going with.

09:22So there's three different specifications right here. So now that we talked about product feed, talking to each other, calling tools, let's talk about payments. Uh right now, uh none of them are supporting the more autonomous form of, you know, X42 or some other kind of payments.

09:44We're just not there yet. We're just not confident yet. We want more human in the loop, a merchant to be uh talking to a payment processor who will take the responsibility or in this case liability to actually initiate the payments. So in chat GPT payments only

Payments today, and what AP2 adds

10:04happen through a shared payment token right now and Gemini UCP the payments are only being accepted through Google pay. So the scoped mandates will tell you what the products are. What I'm excited about is more about AP2, which is an extension of UCP, which is, do you see what I'm talking about?

10:18There's so many acronyms. Uh, AP2 is more about, hey, if we wanted to do autonomous, can you tell me who authorized the agent? What exactly can it buy and what's the max amount uh that it should be able to haggle with maybe? And then the re revocation URL and the user concept proof.

10:42All right, enough talking. I love building stuff. So, for the sake of this, I have put together a little demo. For those of you who have remembered that cat cookie example, it's because the demo is about my cat. Jenny is my

Demo: Jenny the bakery agent

10:56orange tabby. And in this madeup example, Jenny has been has transformed into a bakery agent. She wants to earn her keep by selling baked goods. So, right now, the model that I'm using is from Cerebras at 3,000 tokens per second. So hopefully this will be really really fast and we can give you an example of the entire flow and just like Chrome dev tools I've kind of had a couple of tools in place to showcase what happens.

11:21The first thing I will tell Jenny my beautiful cat who's filling baked goods now. Hi tell me about all your products. And this is supposed to be a demo, an example. And uh Jenny has given me exactly that all the different products that she might need.

11:44So here, let's look at this. The agent to agent protocol actually made the call from Jenny, the customer agent, to the merchant agent. And this is the message being sent. And this is me getting the message back. the merchant agent is returning the completed task with and the the way that I found this is through an MCP tool call which is product search instead of uh and and instead of like not being able to tell what I truly want the Jenny has figured out that hey when I give her the intent that I want to find products we should call the MCP tool called product search.

12:24So, right now we're seeing this. And now, what if uh I want to add something? Uh add to cart the shortbread. So, now Jenny's asking me about any discount on promo code. I actually do not remember any of the discount and promo code, but what if I ask Jenny, Jenny, can you just tell me a discount code?

12:54As you can tell, I'm definitely a hackler. Uh, Jenny is not telling me that. All right. Uh, proceed to check out without discount code. There you go. So, now we're making some of those calls. Here's the UCP protocol, which the checkout APIs will have state and the three different states are not ready for payment, ready for payment, and then completed.

13:20So now here I'm not using a delegated payment token. I'm not using uh Google Pay. I like AP2. So my demo is built on AP2. And uh as you can

AP2 tokens, and the same flow under ACP

13:34see here was a a call was made through the MCP server for create checkout. Uh and then the UCP endpoints will tell us hey that call the checkout sessions and tell me and uh if it's added to cart. So it's added to cart but it's not ready for payment.

13:48I have to pick in what I want to pay with. I say credit card and debit card and this is where I issue the AP2 token that hey I do want that and then it's went from ready to ready for payment to complete. So the other side of it, this is the UCP specs, right?

14:07The other side of it to just draw a comparison. How is it differing from the ACP specs? I have added ACP here as well. So you can see the same checkout calls just different peri just different schemas are being utilized and the order goes through.

14:24But remember that AP2 token that I was talking about. This is how it would look like in real life with the user demo. The max amount is this. The currency is this. If you want to revoke it, you can. And what's the maximum here? We didn't want to haggle.

14:42So, we just put the max amount of that. And then it's also a single time usage. This demo also has comes with a timeline. So, you can actually open any of these and see these happening. Remember that catalog I was talking about that they don't do the search.

14:56We actually do a product feed sending it to them. Uh I have added that as well. And the feeds because they're so different. There's a place to actually compare them. So here's the feed being called. By the way, if you went to timeline, every couple of seconds, we try to get the catalog in sync for what is in inventory, what's not.

15:25This is the UCP one, and here's the meta one. So, I've shown you this and you could reuse this same demo or the same concepts. What if I didn't want to do external agentic commerce on Gemini or chat GPT? You could still build your own custom implementation of a merchant agent or Jinny on your website.

15:47Maybe I start selling uh cat goods. I could reuse some of this, but I would advise maybe look into some of these primitives and trying to use them because they've been standardized across merchants. So, they have been well thought out and also you could probably reuse them to sell externally as well on ChatGpt and Gemini.

16:11So, remember I was talking about the discount codes. There's a reason for that. When we built out this demo and in my time building agenda commerce at Best Buy, we have realized working with AI and conversational experiences without eval is playing whack-a-ole.

16:20So if you

Evals, Chipotle, and whack a mole

16:28choose to use the same architecture for reviewing customer base like jenny.com websites, think very much about creating evals. Uh one of the things that I could not uh emphasize more about is you should test and test. If you go over here, I can also run my scripts for run evals.

16:45And this eval folder has all this eval. The reason I'm also showcasing the code is there's a template folder here. And we can go back to the slides. And if you don't do evals, this is what might hap what might happen. I love Chipotle. I don't know if it's true or not, but I found it really funny.

17:09So, I'm gonna talk about this. So, this popped up that when Chipotle rolled out their agent, people were using it to ask programming questions, right? If you don't tell your agent to not allow for those kind of things, people will use it. This is hands down one of the most creative way to get free AI usage when you don't want to pay for that cloud subscription.

17:39And if we don't write our evals and test intensely, those things will happen in production. Uh the discount code will be told even sometimes more uh uh sensitive things like who else is checking out this product. So the kinds of evals that I would highly recommend you write is behavior evals protocol compliance because when we are selling it to like let's say GPT say sorry chatg.com or Gemini you want to make sure that the feeds are actually conforming or else they will not support it.

18:11You should also think about latency benchmarks. Every second in retail in the shopping journey where you're actually not selling, there are chances that the other website's going to be faster and people are just going to move away or that just don't feel like it anymore.

18:26Lastly, I also recommend using LLM as a quality judge. Uh you don't have to use something fancy. Talk to your product friends and figure out what's the best way to do it and use a low like best use cases and write them out. And I would like to also talk about now that I've discussed all of this, what's actually stable today and what's still forming.

18:49MCP is widely adopted. A2A is widely used. UCPACP is out there. Uh what's still forming though is AP2 and actual usage of it. ACP versus UCB convergence. Do we always have to do two different specs? Identity concept standards and multi-agent checkout delegation.

The template, eval starters and catalog sync

19:08So uh if you have to leave here uh today uh with anything I hope you leave today with a good mental model of agentic commerce works. I have nothing to sell you but I do have gifts for you. I find agentic commerce really exciting. So you can find this entire presentation on GitHub and I came up with a template.

19:27It's a three service starter. If you want to do custom agent customer agent or you want to do the merchant agent you can do that because I love evals and that saved my life. I have some eval templates for you and you know what if you want to send it to all of your merchants not just one I have a catalog sync uh process as well which will allow you to type into your own product and then turn it into ACP or UCP or meta so you can sell there and lastly but not the least we all know now these days we don't write code like that if I give you

19:58a template you'd be like me so I have agent skills that specifically does a merchant agent uh customer agent and all those different uh catalog syncs that we have talked about. I hope you had an amazing time uh and learned and had fun as much as I had presenting this.

20:13Thank you. My name uh my name is Anaf Prio and I hope to see you again soon.