Hello, welcome to the Agent E-Commerce track, and I'm Anil Larmity. I'm a senior solutions architect here at AWS. I'm here to talk to you today about how AWS is innovating and how you can build apps on AWS to support Agent E-Commerce.
So welcome to the session. Just to get you started, let me set the stage with something that you're very familiar with. Imagine that you or your organization is building a news portal like this, right? You're all familiar with something where you're accessing the news content, and then suddenly you hit a paywall, right? So this is where you pull out your wallet, or you try to figure out how to make the payments, set up your credentials, access keys, in the sense that you make a credit card transaction, weekly, monthly, or annual subscription, and then get started to access the content, right? So this is all the content that is behind a paywall.
But what we see now is that much of the traffic that is actually being sent to these portals on the internet is all coming from bots. We see that we're at an inflection point where the bot traffic is more than the human traffic, right? In fact, it's just surpassed that. And 95% of that bot traffic is coming from AI agents.
So essentially, we are also looking at the rise of autonomous agents, right? We all started using LLMs, asking questions, asking for summarization, being able to get help with using them as co-pilots, getting them to do agent work, to do multi-step tasks. And now we are in the phase of autonomous agents, where agents are using the reasoning powers of large language models to complete a task. And completing a task means that it has to go do whatever you're asking it to do. That's where we are in the journey. And we see that by 2027, about a billion agents will be performing tasks, and 60% of the enterprises will already be using agent workflows.
So what happens when agents hit these paywalls that we just saw? When agents hit the paywalls, they stall, they can't operate, and you see those messages that, hey, I cannot access content, right? So at that point, humans get in the loop. They try to enter and put the credit card details or API keys to those transactions for the AI agents. But all of that is manual friction, right? So essentially, bringing in a human in the loop. Autonomous agents actually break at that point, where the friction is now building up.
So now sellers of the content have a couple of options, right? They can block all their bot traffic, but by blocking all the traffic, they lose this AI-powered discovery, they miss these partnership licensing options, and AI also now supports citations, right? So the responses. So they lose all of those powered citations as well if they can't sell the content. Essentially, they lose revenue-generating options.
And if you allow the bots to access the content, what it means is that hundreds of thousands of bots or millions of bots could be hitting your infrastructure, which means that the infrastructure costs will also rise, and you need to be able to support all of that, right? Also, when you allow bots to access content, you lose attribution, the IP itself, right? Because content is now freely available. So both these decisions are probably not a good option. They're not ideal. So there should be another ideal option where you would want to have your AI agents be able to get and pay for the content that they are looking for and monetize on that.
So now we look at the next phase of rising autonomous agents, where agents should be able to transact and discover other agents' resources and essentially make payments, right? So this is the definition of agent e-commerce, where AI agents can essentially discover. It's a form of e-commerce where autonomous agents can discover independently and make those settlements and then access content. So let's look at agent e-commerce, the two sides of agent e-commerce: the buy side and the sell side. So when we talk about the buy side, the agents are making these transactions. And on the sell side, the sellers of the content are trying to monetize the content.
So on the buy side, when you look at things, AI agents want to access this premium paywall content, licensed content. They want to be able to hold wallets, which they do not have the option to do today. And they want to be able to make these microtransactions you just heard in the prior talk as well. But enterprises, when they come to this point, want more guardrails, and they do not want agents to go on a spending spree. Think about it, right? Would you allow your AI agents to get a handle on your wallets or credit cards to be able to do that, right? Transactions where they could go rogue as well, right? So that's what the buyer side is looking at.
And on the seller side, there are, again, billions of transactions that will be happening with these AI bots. So sellers really want to be able to understand what kinds of bots are operating, what kinds of transactions they're making, and really do this at the edge. The sellers don't want to change their entire infrastructure and origins where the content is sitting. They want to be able to do this at the edge without changing much of this, right? So there is, again, one common thing here on the buyer side and the seller side, which is a standardized approach or a protocol to be able to solve for this machine-to-machine payments at the edge.
So bottom line, buyers are saying that they want their agents to be able to pay for content and not have humans approving this. And then the sellers are saying that they want to be able to earn from the AI traffic. So bottom line, the subscription model is going to change from humans in the loop to becoming humans on the loop or out of the loop. And that's what we are building toward.
The traditional one-size-fits model does not work anymore because of the fact that, again, we look at that in the next slide, where the transaction costs will not really work, right? All of this needs to be happening at real-time speed, and the pay-per-use and pay-per-execution model is what the future is going to look like.
So if you're a seller, you would have come across this, right? There is a 25-cent minimum transaction fee as well as 2.5% on top of that. And all of these microtransactions are in a cent, sub-cent, or micro-cents, is what we are calling them. So if you add 25 cents on top of that, it's essentially 250 times what they are essentially paying for. So all of this model does not work.
And while we are trying to solve for that, a very brief history of this is every HTTP call essentially responds back. There is a response for that. You've seen 200 status codes, 404, and 301. These are all status codes that you're familiar with. And then there is one status code, which is 402, which has not been used. It was reserved for payment required. And now finally Coinbase has introduced this as transactions over 402, which is also called X402, where the protocol talks about how you can do machine-to-machine transactions using this protocol, right?
So we'll take a closer look at that. But what happens within the protocol is, if you look at this flow chart here, a client makes a request to the server, and then the server responds back with the payment required. The client then figures out what is the payment method that it wants to operate, and then it sends the payment authorization to the server. The server then utilizes a facilitator to complete the verification and also utilizes the same facilitator to complete the transaction. And once the settlement is completed on-chain, essentially the server will then respond back with the content, right? So this is what's happening under the X402 protocol.
I thought I'll pick one of the protocols and just explain this to you. But why this is compelling is, essentially, there are no protocol fees, or the fees that a consumer is paying for these micro-send transactions, and the merchant is paying very nominal gas fees. Again, there is zero wait time. This is happening at the speed of the internet, and there is no friction. There are no API keys to set up, no subscriptions, and the payment is essentially the credential to be able to get the content. So there is no centralization. X402 can be extended as well, and you can implement it, and there are no restrictions as well.
So some key milestones here are, it was introduced last year, May 2025. X402 is now part of the Linux Foundation under open governance, and it's backed by Coinbase, AWS, Google, Stripe, Anthropic, Cloudflare, and Circle, right? So many more organizations are supporting that. So from Amazon, we have also released AgentCorp payments under the Bedrock suite, where agents will be able to make payments, and we'll go into some of the details here. So let's talk about the buyer side here and what is involved, right? So we understood from the developers that they really want to be able to get these agents and the payment is essentially the credential
to be able to get the content. So there is no centralization. It's X402 can be extended as well, and you can implement it, and there are no restrictions as well. So some key milestones here are, it was introduced last year, May 2025. X402 is now part of the Linux Foundation under open governance, and it's backed by Coinbase, AWS, Google, Stripe, Anthropic, Cloudflare, and Circle, right? So many more folks are supporting that, organizations in there. So from Amazon, we have also released AgentCorp payments under the Bedrock suite, where agents will be able to make payments, and we'll go into some of the details here. So let's talk about the buyer side here
and what is involved, right? So we understood from the developers that they really want to be able to get these agents to have wallet support. They want to be able to have real-time settlement, have the budget and guardrails, which enterprises really want, and observability throughout the stack, where they would want to have the full stack trace of everything that's happening under the hood. So I'm excited to share with you that we've launched AgentCorp payments, and this is a service that allows AI agents to autonomously discover, authorize, and execute payments with a few lines of code. Now we've launched this in partnership with Coinbase and Stripe,
where you can bring wallets from Coinbase and Stripe Preview to be able to do these operations, and we'll go into some of the details. But the core capabilities to start with are wallet support, where you can bring the wallets from Coinbase and Stripe. You're able to orchestrate the payments using payment connectors, and today we support X402 with many more protocols that are in the pipeline. The service is designed to be protocol agnostic, so as new protocols emerge, we are going to be adding support for those protocols as well. And the settlement is going to be instantaneous, instant essentially, and the payment limits can be set,
which is the most important thing that we spoke about, where enterprises are looking to put some payment limits and guardrails on how these transactions can operate. So observability is built in, and essentially all of this operates with security as the layer that is operating the whole model. So with that, let's look at some of these details on how the payments limit can be set up. So you can create payment sessions where you can programmatically set the maximum amount of value that can be used for transactions, or you can also set expiry time in minutes. Think where you are able to set that the agent can actually spend maybe $5 in 30 days or 60 days, right?
So that's the operation model that you can set with many more details that are available. I'm only going over a few features, but let's look at what happens on the buyer side. When the user is asking an agent to make a particular request for resources, right? So the agent completes the request by accessing tools, MPP servers, other resources as well. So at that point in time, if the agent is looking at, it finds that there is a response from one of the tool calls or requests with a 402, AgentCorp payments is going to handle the request to complete the transaction, and then let the AI agent know that essentially the settlement happened,
and the AI agent will be able to respond back to the users. So in this process, when the wallets are, wallet support is imported, the secret keys that you use to import the wallets actually are stored in a secure token wallet that is secured by KMS, where that's there. So essentially the agent does not have access to the private keys. This is most important to note. And the next thing is that AgentCorp payments is also integrated through a gateway, which is also part of, which is another service that we have to MCP-fy your internal APIs. Through AgentCorp gateway, the AgentCorp payments can get access to discovery service in Coinbase,
where there are 10,000 plus endpoints that are available to transact. And then, again, there is a per-session budget that we just discussed as well. So there is a decoupling of agent infrastructure and the payment infrastructure by design, where the agent can operate in its own loop, and whenever it sees the payment, the payment connectors, orchestration, payment limits, and integration with third-party wallets can happen, right? So it's important to decouple them because, again, skills can be poisoned. Inputs for the agents can also be poisoned by inputs as well, right? So where malicious actors could try to do that. So by decoupling and making this by design,
agents can essentially have a secure path for these transactions. And payments do not touch the undeterministic path, but this is more on a deterministic layer as well. So why this is important is that, again, the code of the agents does not have to change. You can bring your own model frameworks, and then the payment itself can flow through in the payment layer itself. So, again, the controls, the policy spending controls can be outside of the payment stack itself. And, again, this is built to be protocol agnostic. So this is one of the console screens where it shows how you can import the payment connector, and it shows that you can select the Coinbase wallet
and the Stripe Preview wallet from the console. And this is a demo in action where we are showing how a secure resource can be accessed. Now, in this case, the AI agent is essentially discovering that there is a secure resource. The AgentCorp payments is kicking in, and then it's completing the transaction by utilizing the wallet that is already integrated, and the transaction completes. Now, this is on the buyer side. Now, let's look at the seller side to understand what's happening, right? So, again, there is a lot of bot activity that's happening. We have released under the AWS Web Application Firewall a feature where we have bot detection in place.
Today, we detect over 650 different types of bots, types of bots like Perplexity Bot, GPD Bot, Cloud Bot, again, Google Bots, right? So there are so many bots that are out there. So we're able to detect, also understand the intent of these bots. So why are these bots accessing the content? Are they accessing the content to train their models? Are they doing it because they have to respond back to an intent where they're responding for a rag search? So we're able to identify the intent. We're also able to verify the bots and identify them by a signature. So we are able to say, hey, this is a verified bot. So maybe you have built a relation with one of these organizations,
and this verification will allow you to have different pricing for the organizations that are already verified. So we'll look at that in a second. So there's also a real-time traffic analysis that allows more to be customized. And I'm also happy to share with you today that we announced WAF AI traffic monetization. This is a service that allows you to monetize based on the content, based on how you can measure, verify, and monetize based on the AI traffic that is hitting your endpoints. Now, if you might be familiar with CloudFront, which is our content distribution network, you can add a web application firewall at that point,
and essentially you can start monetizing right away. And based on a few clicks, you can do that, again, using infrastructure as code as well. Now, I also spoke about a gateway service that allows you to expose your AI endpoints that are internal and MCP-fy them. So the same web application firewalls can be used there. So your internal APIs can be MCP-fied, and then you can start monetizing as well. So what happens during monetization? The AI agent, AI bot essentially requests some content. The bot context understands what kinds of bots, is detecting it. It's able to detect the bot. It's able to categorize and understand the intent of the bot, as we discussed earlier,
and verify and check what kind of bot is available. So then we are able to monetize using the X402, and the publishers get paid as well. So important to note is that, again, there is no SDK change, no changes at the origin. Publishers keep 100% of the revenue as well, and again, there are no transaction fees or subscription fees. So this supports X402, and we are adding support for more protocols as well. A few dimensions on how you can start monetizing. Think you have separate paths. So a slash blog, in this case, can be charging for a different rate than a slash research or maybe an API endpoint itself. And you know the identity of these bots.
Again, if you make some kind of relationship with the bots, companies, organizations, maybe you make a relation with Anthropic, then you can essentially have different pricing for those bots versus different unverified bots. Right, so think of that option. And then you can also set different pricing for intent as well. If somebody is coming here, if a bot is accessing the content for, again, training, you can charge a different rate than what it's doing for a search as well. So again, these are different WAF rules. They can be in a combination of and or, or, or, and then you can access that. and we are adding support for more protocols as well.
A few dimensions on how you can start monetizing. Think you have separate paths. So a slash blog, in this case, can be charging for a different rate than a slash research or maybe an API endpoint itself. And the identity of these bots. Again, if you make some kind of a relationship with the bots, companies, organizations, maybe you make a relation with Anthropic, then you can essentially have a different pricing for those bots versus different unverified bots. Right, so think of that option. And then you can also set different pricing for intent as well. If somebody is coming here, if a bot is axing the content for, again, training, you can charge a different grade
than what it's doing for a search as well. So again, these are different WAF rules. They can be in a combination of and or, or, or, and then you can access that. So this is how the re-imagined flow would look like where you're allowing the AI agents or essentially verified bots and unverified bots to have different pricing and humans to have different pricing. Some cases you want to have humans to access the content freely. Some cases, again, the humans could be charged, where the bots could be charged differently as well. So this is how you can reimagine the price. So again, there are some dashboards that show the revenue numbers
and how you're able to aggregate by different bots and figure out what kind of revenue model you want to operate. And it also shows what is the path that's being accessed by these bots. So again, what is currently everyone using Agent eCommerce for? They are using Agent eCommerce to run, again, LLM inference, getting compute, web scraping. They're creating research agents to be able to serve the requests, and agent to agent as well. We see MCPs also being monetized now. Again, this is the last 12 months of traffic from, again, what we are seeing on Coinbase agentic market, where you're seeing that a $50 million volume transaction happened over 170 million transactions.
The average settlement time is 200 milliseconds on base with about a tenth of a cent as cost per transaction. So I spoke to you about Agent Core payments, which is one of the parts of the bigger ecosystem, Agent, Bedrock Agent Core, where you can essentially bring your own model, you can bring your own framework, and you can start building AI agents. You can add context by adding memory. You can bring, again, your own managed knowledge bases. You can add web search capabilities to the agents. You can MCPfy your internal APIs, and then you can have many more features, like being able to run evaluation on how your agents are performing. So again, you can use runtime,
which is a Bedrock Agent Core runtime, where you can bring your own agent application and serve at scale, and every request will have its own isolated micro virtual machine that is running to serve the requests. So that's it from my side here today. Thank you, and I hope you have a nice day. Thank you. can be outside of the payment stack itself. And, again, this is built to be protocol agnostic. So this is one of the console screens where it shows how you can import the payment connector, and it shows that, you know, you can select the Coinbase wallet and the Stripe Preview wallet from the console. And this is a demo in action
where we are showing how a secure resource can be accessed. Now, in this case, the AI agent is essentially making a, you know, discovering that there is a secure resource. The agent code payments is kicking in, and then it's completing the transaction by utilizing the wallet that is already integrated, and the transaction completes. Now, this is on the buyer side. Now, let's look at the seller side to understand what's happening, right? So, again, there is a lot of bot activity that's happening. We have released under the AWS Web Application Firewall a feature where we have bot detection in place. Today, we detect over 650 different types of bots,
things of bots like Perplexity Bot, GPD Bot, Cloud Bot, you know, again, Google Bots, right? So there are so many bots that are out there. So we're able to detect, also understand the intent of these bots. So why are these bots accessing the content? Are they accessing the content to train their models? Are they doing it because they have to respond back to an intent where they're responding for a rag search? So we're able to identify the intent. We're also able to verify the bots and identify them by a signature. So we are able to say, hey, this is a verified bot. So maybe you have built a relation with one of these organizations, and these verification will allow you
to have a different pricing for the organizations that are already verified. So we'll look at that in a second. So there's also a real-time traffic analysis that allows more to be customized. And I'm also happy to share with you today that we announced WAF AI traffic monetization. This is a service that allows you to monetize based on the content that, you know, based on how you can measure, verify, and monetize based on the AI traffic that is hitting your endpoints. Now, if you might be familiar with CloudFront, which is our content distribution network, you can add a web application firewall at that point, and essentially you can start monetizing right away.
And based on a few clicks, you can do that, again, using infrastructure as code as well. Now, I also spoke about a gateway service that allows you to expose your AI endpoints that are internal and MCP-fy them. So the same web application firewalls can be used there. So your internal APIs can be MCP-fied, and then you can start monetizing as well. So what happens during monetization? The AI agent, AI bot essentially requests for some content. The bot context understands what kinds of bots is detecting it. It's able to detect the bot. It's able to categorize and understand the intent of the bot, as we discussed earlier, and verify and check what kind of bot is available.
So then we are able to monetize using the X402, and the publishers get paid as well. So important to note is that, again, there is no SDK change, no changes at the origin. Publishers keep 100% of the revenue as well, and again, there is no transaction fees or subscription fees. So this supports X402, and we are adding support for more protocols as well. A few dimensions on how you can start monetizing. Think you have separate paths. So a slash blog, in this case, can be charging for a different rate than a slash research or maybe an API endpoint itself. And you know the identity of these bots. Again, if you make some kind of a relationship
with the bots, companies, organizations, maybe you make a relation with Anthropic, then you can essentially have a different pricing for those bots versus different unverified bots. Right, so think of that option. And then you can also set different pricing for intent as well. If somebody is coming here, if a bot is axing the content for, again, training, you can charge a different grade than what it's doing for a search as well. So again, these are different WAF rules. They can be in a combination of and or, or, or, and then you can access that. So this is how the re-imagined flow would look like where you're allowing the AI agents or essentially verified bots
and unverified bots to have different pricing and humans to have different pricing. Some cases you want to have humans to access the content freely. Some cases, again, the humans could be charged, where the bots could be charged differently as well. So this is how you can reimagine the price. So again, there is some dashboards that show the revenue numbers and how you're able to aggregate by different bots and figure out what kind of revenue model you want to operate. And it also shows what is the path that's being accessed by these bots. So again, what is currently everyone using Agent eCommerce for? They are using Agent eCommerce to run, again,
LLM inference, getting compute, web scraping. They're creating research agents to be able to serve the requests, and agent to agent as well. We see MCPs also being monetized now. Again, this is the last 12 months of traffic from, again, what we are seeing on Coinbase agentic market, where you're seeing that a $50 million volume transaction happened over 170 million transactions. The average settlement time is 200 milliseconds on base with about a tenth of a cent as cost per transaction. So I spoke to you about Agent Core payments, which is one of the parts of the bigger ecosystem, Agent, Bedrock Agent Core, where you can essentially bring your own model,
you can bring your own framework, and you can start building AI agents. You can add context by adding memory. You can bring, again, your own managed knowledge bases. You can add web search capabilities to the agents. You can MCPfy your internal APIs, and then you can have many more features, like being able to run evaluation on how your agents are performing. So again, you can use runtime, which is a Bedrock Agent Core runtime, where you can bring your own agent application and serve at scale, and every request will have its own isolated micro virtual machine that is running to serve the requests. So that's it from my side here today.
Thank you, and I hope you have a nice day. Thank you.