The Startup Ideas Podcast - Amazon & Delta Are Blocking AI Agents. How to Win Anyway.

Episode Date: October 21, 2026

In this solo episode, I break down the war on AI agents, from Amazon blocking Meta's Muse to Delta treating outside booking agents as unauthorized automated activity to Yelp charging agents for access. I share how I think about an agent, as a customer with a credit card who reads your site instead of looking at it, and the ladder of trust that runs from research to prepare to pay. Then I walk through each big player's bet, my nine stages for how this war plays out, five moves any business can make this week, and the startup ideas I see between agents and the businesses they buy from. If you run a store or build products, this episode gives you a map for selling to customers who send an agent to shop for them. Timestamps 00:00 – Intro 02:09 –The Three Camps 05:08 – How I Define an Agent 06:52 – The Ladder of Trust 08:54 – Real Reasons to Block Agents 09:48 – What Agents Skip 12:57 – How Each Company is Playing it 17:06 – Lessons From Mint and Plaid 17:49 – The 9 stages of The War on Agents 21:56 – What to do this week 24:39 – Startup Idea 1: Agent Readiness 25:48 – Startup Idea 2: Plaid for One Vertical 26:22 – Startup Idea 3: Insurance for Agents 26:57 – Startup Idea 4: Analytics for Agent Traffic 27:26 – Startup Idea 4: Agent Specialist 28:05 – Closing Thoughts Key Points Agents read product details, prices, shipping, and return policies, so plain-text facts on the page decide whether they pick you. Agents climb a ladder of trust from research to prepare to pay, and the whole fight happens on the pay rung. Many blocks protect the ad and upsell revenue that agents skip on the way to the sale. The biggest stores block, the biggest audiences open the door, and payment and security companies want the toll booth. I expect a Plaid moment for agent identity and payments, followed later by ads that buy an agent's first recommendation. Vertical Plaids, agent insurance, and agent analytics are open lanes for startups right now. The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/

What's the deal with this war on AI agents? I sound like Jerry Seinfeld there, but I'm being serious. A few weeks ago, Amazon started blocking Meta's new AI agent Mews from its website. Delta says it works with zero outside AI agents for booking flights, and it's treating them as unauthorized automated activity. Yelp is now saying agents can come in, but only if they pay. So we're kind of at this strange moment. The biggest tech companies in the world spent years, two years, telling us AI agents are going to do everything for us,

book the trip and buy the groceries and handle the errands. And now the places where those errands actually happen are starting to shut the door. I've been calling it the war on agents and a few people are catching on. And I think it's one of the most important things happening on the internet right now because whoever wins this fight decides how you and I are going to, going to buy things for the next 10 plus years. So here's the plan. I'm going to show you what's actually happening. I'm going to give you a better way to think about an agent. I'm going to walk through how each big company is playing this and tell you how I think the war plays out stage by stage. Now, if you're new to the channel, my name is Greg Eisenberg. I've been advised of companies like

TikTok, Reddit, and I'm a co-founder in LCA, which advises the world's biggest companies on things like AI. So I know a few things about that. And I want to talk also by the end of this episode about where I think fun opportunities are, profitable opportunities are, because every time the internet had a fight like this, a few people got very wealthy, building the bridge between two sides. So that's kind of why I wanted to take a second and actually do this episode. There's one stage in particular that I think almost nobody is paying attention to, and I'll get to it by the end of the episode. All right, let's chat war on AI agents. So what's the lay of the land here? Because the headlines make it sound like everyone is blocking AI agents,

and that's not exactly what's going on. So on the blocking side, Amazon started blocking Meta's Muse agent from the store around September. And if you ask Muse to buy something on Amazon, it can't browse the catalog or it can't check out. Like, you can't buy anything. This is actually familiar territory for Amazon, by the way. I remember last year, I think they went after perplexity shopping agent. Now, Delta says it works with zero outside agents for shopping or booking, and that it's protecting customers from unauthorized automated activity. That's what they keep saying.

And then there's also Yelp. So Yelp lets agents in only if they paid through their data licensing program. So they'll allow you in, but you got to pay price. Then there's this middle group. I actually think this middle group is the most interesting one. So Zillow says agents acting for real customers are welcome, but it's anti-scraping tools sometimes block them anyways. And then you also have Walmart.

So Walmart had customers saying Muse kept failing on its site. And Walmart said that was an accident. So Walmart is actually a Mews partner. By the way, I have no affiliation with Mews. I just think it's interesting to comment on here and interesting to find out where the opportunities are. So that's a detail worth sitting on, by the way, around Walmart because a lot of these blocks are a security system doing what it was actually built to do.

So there's defenses that we've built over the last many years. designed to stop scrapers, and a lot of websites are just not able to tell the difference between the scraper and your personal assistant. And on the welcoming side, META is working with Walmart, they're working with Stripe, and a handful of others on an open standard for how agents and businesses talk to each other. Shopify, you may have seen this, built away for merchants to sell through AI channels, and Visa and MasterCard are both building ways to recognize trusted agents. Meta's line on this was basically, if you turn away a personal agent, you're turning away a customer, and I understand where they're coming from. So you've got three

camps, right? You've got the block camp, you've got the accidentally block camp, and then you've got the come on in, welcome camp. Which camp a company lands really tells you a lot about how it makes money, and we're going to get into that. But first, I wanted to talk about what an agent actually is explained clearly, because I think the way most people picture it is wrong. So a lot of people, when they picture an AI agent, they picture this robot just clicking around a website. And that's how it works today, but it's actually the wrong way of thinking about it. So how I think about it is an agent is a customer with a credit card and no eyes. By the way, I know there's going to be people who are going to say, I'm thinking in the future, this is never going to happen.

This is the arrow of progress where things are going. You're going to have computers, virtual computers, with a credit card, who have intelligence and no eyes. And what I mean by no eyes is it can't see your homepage video that you spent $100,000 building. It doesn't notice your banner, your pop up on the countdown timer on your sale. I mean, it reads your product details, your prices, your shipping costs. It will read your return policy, but that's basically it. So everything you spent years designing for human eyes is mostly invisible to it. Okay, let's try to make this real so it clicks in your head.

Say you need a replacement filter for your air purifier. I'm staring at mine right there, so that's why it's top of mind. You tell your agent the model number and you say, find the right one, compare the delivered price, and check with me before you buy anything over $50. Notice what you just did. You gave it a goal. You gave it rules and you gave it a limit. That's the second thing to really understand about agents. They work on a ladder of trust.

So the first rung is research. So find options, compare them, tell me what you found. The second rung is prepare. Put it in the cart, hold the seats, get me a quote. And the third rung is pay. Actually, go and spend my money. So almost every company that I've seen is fine with the first rung because that's basically search. The whole war is happening on the third rung.

And I think a lot of businesses are making a mistake by blocking all three. when the real risk only potentially lives within one. The third thing is that an agent is the most disciplined shopper you'll ever meet, right? It never gets tired halfway through checkout. It doesn't impulse buy the next thing next to the register. You know, it finds the shipping fee you hit on the last page. And it compares five suppliers in the time it takes you to open one tab. And the thing that I find underrated is agents have this digital.

different kind of loyalty. So you and I are loyal to brands because they make us feel a certain way. Maybe we're nostalgic for it. Maybe when we were a kid, we use that brand. And it's, sometimes it's honestly subconscious. An agent is loyal to whatever worked last time. So if your site got the order right with the right part delivered when you said, the agent's going to come back to you and it's going to be in the memory of that agent. But if it got a stuck on your checkout. It's going to go somewhere else probably and you just won't know it happen. So think about it that way. A customer with no eyes, a ladder of trust, and loyalty that goes to whatever works. And once you see agents like that, it becomes pretty obvious why some

companies are scared. And how I think about it is they're leading from a place of fear. So why are companies really blocking agents. And I want to be fair here because there are real reasons to block AI agents. And I want to give like a balanced pros and cons in this episode of the podcast. So if an agent orders 50 orders instead of five, somebody has to fix it. And the customer is going to obviously call the store or company. And they're not going to call the AI company. So if an airline can't prove that you actually authorized a booking, that's a fraud problem. And plenty of the traffic calling itself an agent is really just a scraper polling prices or training data. So when Delta says it's protecting customers, I do believe that's part of it, even though it sounds kind of lame. I do

believe that it is part of it. So I'll give them credit there. Now, Amazon makes tens of billions a year from ads. And I think it's like one of the, it's just a very profitable business line for them and it's growing a lot. So a lot of that is sponsored products that are sitting at the top of your search results. So an agent is going to skip past those

and it's going to pick the product that matches what you ask for. Airlines make a fortune on upsells. Things like seat upgrades, bags, even the credit card offer at checkout, but an agent is going to book you the cheapest nonstop. It's not going to click any of that, right? So for a lot of these companies, the agent isn't stealing the sale, so to speak. They're still getting the sale, but the agent is taking away everything they used to do on the way

to the sale, and sometimes those happen to be the highest margin parts of those businesses. And I think, you know, there's a saying, you know, follow the money. You know, this is sort of following the money, and that's why they are blocking it. And I think there's a bigger fear underneath that one. You know, if everyone shops through an agent's built by, you know, meta or OpenAI or whoever, then that company sits between the store and the customer. And we've seen that movie before, you know. So I think a lot of us, you know,

or a lot of companies are just looking at that and they're like, that we don't feel comfortable in a world where meta or OpenAI or whoever is between me and my customer. The part that's kind of funny, though, is the same companies blocking outside agents are building their own. So Amazon has Rufus. Delta, you know, did some research for this episode. They have their own AI concierge. So the position isn't really agents are bad. It's agents are great as long as they work for us and we own them.

This one made me question something I've said for years, which is own the customer relationship. I still believe in it. But there's a version of that idea that turns into make the customer deal with me the way I prefer. And if I've trusted an assistant to handle an errand, making me take the errand back isn't a better relationship and it's not a better experience for the customer. It's just more work for the customer. So I don't think that's very good. So the question, you know, I'd ask any business is, what part of your relationship with customers comes from being useful and what part comes from making it hard to go anywhere else. It is an uncomfortable question if you've gotten really, you know,

if you've gotten really good at the second one, but it's something that you should be asking yourself. Now, let's look at how each of the big players are actually approaching this, because they're all going to make different bets, and I think we can learn something from that. So how are the big companies thinking about this war on AI agents? I want to go player by player, and give me my best read on what's going on. Because once you see everyone's bet side by side, the whole thing I think is going to make a lot more sense. So Amazon is a walled garden.

It blocks outside agents and it builds its own. Its bet is that Amazon is going to be the destination and it's the everything commerce app. And it wants you shopping with Amazon's assistant inside Amazon store looking at Amazon ads. Meta's the opposite. Muse sits inside apps billions of people already use, and meta wants every business to let agents in.

It's working on an open standard with Walmart, Stripe, and others. But of course, Meta wants the door open because meta owns the agents walking through it. And also, if you just think about the history, of Meta. Mark Zuckerberg has talked about this. I was just recently watching the episode with Alex Wang and Mark Zuckerberg on the Cheeky Pint podcast. And he talks about basically how, you know, the era of Apple and Android, you know, because of those big ecosystems, Meta hasn't been able to do what it's want to do.

So they want to control their destiny, and that's why they're doing that. Walmart's a pretty interesting one. It is Amazon's biggest competitor, and it went the other way and partnered with Mews. So when you're number two, opening the door to a new channel is how you take share from number one,

so I understand strategically why they did that. Open AI's bet is to put the checkout right inside the conversation. So you can already buy, from Etsy and Shopify stores without leaving chat GPT. I think where they're going is just to own the moment you decide to buy. Google already owns the biggest toll booth on the internet, which is Google Search, an incredible product. So it's building agents and payments into Gemini while trying to protect an ads business.

That depends on clicking things. So it'll be interesting to see how that plays out. Shopify is playing arms dealer. It's agentic storefronts, let merchants plug in once and show up across AI channels. So basically, Shopify is saying, I don't care what agent wins. As long as merchants are in all of them, and if they're able to basically increase commerce on the internet, I think Shopify is pretty happy. Visa and MasterCard are building ways for stores to recognize, trusted agents. And they're working with Cloudfair on it. I've done an episode on Cloudflare and some

of their agent products. You can go search that if you want to watch that. Their bed is basically whoever wins, the money's still going to move through them. And Cloudflare, you know, sits on the door itself. You know, it protects a huge chunk of the internet. And it's a reason a lot of the agents are getting blocked by accident. And it's also building ways for agents to pay their way in. I think it wants to be the bouncer, basically. So there's a pattern here, right? You're noticing it. The companies with the biggest stores are blocking.

The companies with the biggest audiences are just opening the door. And the payment and security companies want to be the toll booth no matter who wins. And, you know, if you've watched the internet for a while, you know, you've been paying attention. I think you know how a lot of these fights usually end. So how do I think the war on agents probably play out? You know, I've seen this fight before. I remember mint.com, some of you probably as well, which allowed you to basically log in to your bank account

and sort of get data and analytics on it. Banks hated it originally. I don't know if you remember that. Banks originally fought it really hard. And then a company called Plaid showed up, and they built a proper connection between the apps and the banks, and they became worth billions. The banks went from blocking to signing deals to charging for access.

So here's how I think the war on agents plays out, and I think there's going to be nine stages. I tweeted this. I got some people agreeing with me, some people who disagree with me. So let me know where you set. Stage one is where companies block agents because agents are skipping ads and the upsells. And I think that's where we are today. These are early days for personal agent platforms.

They're just coming out. They're going to have to iterate to scale and product market fit and all that stuff. So it's early days. Stage two is agents learn to act more like humans to get through. And it turns really into this arm race. So every block gets a workaround, and every workaround gets a new block. Stage three is where the market is going to split. So companies that make money on sales let agents in because they want orders.

They're making money. Companies that make money on data charge for access. That's why companies like Yelp are charging for access. And the giants build their own agent, personal agent platforms. like what we're seeing with Amazon. Stage four is where a trusted middle layer for agent identity and payments becomes the standard way in. And that's going to be the plaid moment for personal agents.

And you can see Visa, MasterCard, and Cloudflare already racing for it. Who wins? We will see. Stage five is where companies that welcome agents start winning customers from the ones that block them. So if one airline lets your agent book in 10 seconds and the other makes you do it yourself, you know which one your agent is going to pick next time, like we talked about earlier. And stage seven, everyone just starts competing for agents. So better prices for agents, faster checkouts for agents, product pages written for agents, SEOBA for agents.

There's going to be tons of just help that companies are going to need for this stage it's going to what's the equipment of a billboard for an AI agent lots to unpack there and then stage eight

is when the ads come back and companies are going to pay to be the first thing your agent recommends and this is the stage I mentioned at the beginning of this episode and the one that I don't think

a lot of people are talking about the whole reason companies block agents is that agents skip ads. But advertising isn't going to disappear. It's just going to evolve into something else. We've seen that time and time again. Just look at the history of advertising from print ads, TV ads, radio ads, digital marketing, etc., etc., right? I think that Google has actually tested ads inside AI search. I know ChatGBT has, right?

So that's the beginning of stage eight. You're trying to see companies, like, test it. I can confirm that I'm pretty sure, yeah, I can confirm that chat Cheptee is doing it. And I think I saw an article that Google's doing it. But for sure, chat Chb-T. Stage 9, by the way, the war ends. And selling the agents are just going to become normal as normal as having a website. In a few years, a business that can't be bought by an agent.

is going to look like a business without a website looked in, I don't know, 2005. Now, I don't know if these stages are going to happen in such a neat order. Sometimes they just end up blending in together. Some companies are actually at stage six while others are stage one. But if you're trying to figure out where to build or how to make sense of all this stuff, I think this map tells you a lot. So what would I do if I ran a business, like what would I do this week based on all the things that I'm telling you about AI agents and where, you know, where this is going.

And by the way, none of this requires like a huge tech project. I would just, you know, first run an agent test. So go into your support inbox, pull 10, you know, real buying questions. Does this fit the older model? Can I get 12 of these by Friday? Give each one to two or three different agents and watch what happens. where does it get stuck? That exercise is going to teach you more than any report or anything,

and you kind of just have to do that if you believe that personal agents are going to become a thing. Second, check whether you're blocking agents by accident. Remember Walmart, a lot of businesses have bought protection turned on. They set it up years ago for an internet that was very different than the internet of today, and they just haven't looked at it again. So go look at it and make that decision on purpose. Third, make your offer readable. If your compatibility info lives in a PDF or your delivery date only shows up on the last checkout screen, an agent's not going to be able to easily use that.

So put the important stuff in plain text on the page, total price, delivery date, what it works with, how returns work. Fourth is you're going to want to make your advantage readable too. So say your product costs more, but you deliver tomorrow and the cheaper person takes three weeks. You know, if the agent can only see the price, you're going to lose, right? But if they can see the delivery date, you might win. Agents won't turn every market into the race to the cheapest because, you know, their bosses are still people, and some people want reliability, they want speed. So, you know, the important thing is to make sure the agent can see why you're the right answer

because you're probably not optimizing it today. And then fifth, just decide what agents can do on each rung of the latter. You know, let them research freely, let them build a card or request a quote, then add an extra check before money moves, especially on big orders. So you don't need to choose between block everything and allow everything. You can kind of be in the middle depending on what your comfort level is right now. Okay, this podcast is called The Startup Ideas podcast. So it wouldn't be an episode without telling you where I think some of the startup opportunities are

and where there's opportunities. and I've been waiting honestly, patiently waiting to tell you this. So every stage of this war on agents creates a business. So what are the ones that I think are interesting? One is agent readiness for one industry. Every platform shift creates confusion. And confusion is the most profitable product there is. In 1998, people got wealthy building websites.

for local businesses. Right now, every store owner is hearing about AI shopping, personal agents, and has no idea what to do. So pick an industry, you know, dental supply, replacement parts, like something niche. Record three examples of an agent picking the wrong item on a supplier site and then show them what's missing and fix it. Do that by hand 10 times and you'll know exactly what software to build. Two, we talked about plaid, go and build the plaid for one vertical.

The big general layer for agent identity and payments is going to be won, in my opinion, by someone like Visa or MasterCard or Cloudflare. But every industry has its own messy systems. So booking a contractor, ordering restaurant, surprise, scheduling a vet appointment, someone builds the trusted connection between agents and the business. and the businesses in that industry and owns it, and I think there's a bunch of opportunity there. Three, and I've tweeted about this before, and I think some of these ideas have been on ideabrowser.com. Well, insurance for agent mistakes.

A big reason companies block agents is that nobody knows who pays when the agent orders the wrong thing. So whoever can guarantee that the order unlocks the door for every nervous business, well, that's a stage four and a stage six business that I think would do pretty well if you can nail that.

Four is analytics for agent traffic. Most businesses today can't tell an agent visit from a human one. So they don't know which agents are showing up, what are they asking for, why are they leaving? I think that someone is going to build the Google Analytics for agents. and every business is going to need it in a world where personal agents is a big thing.

Five, be the specialist. If you're the supplier who gets parts to independent restaurants and the next day, agents make that easier to find because the customer can ask for exactly that. You know, agents reward businesses that are excellent at one specific thing. So if you can do that, there's opportunity. You know, one thing I'd stay away from is any business whose edge is a trick to get mentioned by AI. You know, tricks stop working. I mean, they work for a bit and then they stop working. So having the right product in delivering it when you said you would is a lot harder to copy.

So where do I land when it comes to the war? on AI agents. If you've made it to the end of the episode, first of all, thank you for being here, like comment and subscribe for more. But, you know, if you're interested in where I think we've landed, I think the most of the blanket blocking that we're seeing is coming from a place of fear. Now, some of it is fair fear. Like, I understand why it's coming. You know, if an agent can't prove you authorize a purchase, blocking it for now might be the responsible thing. But I'd want that block to come with a plan for what it would take to let agents in safely because this isn't safe yet is a problem you could fix.

We don't want customers using their own agents is a reason to never change. So the test I keep coming back to is this. Are you protecting the customer or are you protecting a habit you like the customer to keep? And can you be honest with that? And I don't think welcoming agents mean giving up your business. You can build your own agent. You can set the rules for what outside agents are allowed to do. You can say no to orders and you can, you know, if those orders are maybe not profitable or you can't fill it safely. And you can still create a way for other people's agents to bring you customers. So I don't think it's that. I think the pie is going to get bigger. And, um, And because of that, I don't think you're going to give up, or I don't think people are going to give up their business. If you remember one thing from this episode,

remember the customer has no eyes now. You know, it's already shopping. It's reading instead of looking. It's loyal to whatever worked last time. And it's going to tell your customers where to spend their money. And we're talking about billions and billions of dollars, right? the businesses that make themselves easy for it to understand

are going to pick up customers, their competitors probably aren't going to see. So pick one buying question this week, give it to an agent, watch where it gets stuck on your site, then do the same thing on your competitor's site, and you're going to learn a lot. At the very least, you're going to learn a lot.

Let me know in the comments which stage you think you're in, which of these opportunities you go after, say hi, just show me you're alive. I read every single comment. I'm rooting for you. I can't wait to see what you build.

It's an incredible time to be building. I love, love, love making these episodes of the podcast. I love doing this podcast. I'm grateful that I get to do it, that you listen. It's just a dream for me. I'm sharing everything I can in real time, thoughts, half-baked ideas, just everything. And I hope you feel,

how much fun I'm having, how much I care about you, and how much I care to see you win. I'll see you on the next one and have a creative day.