The Next $100B Market: Selling To AI Agents

0:00There's a really big shift happening right now on the internet and I don't think people are talking about it. For the longest time, the user of the internet were human beings. They were us, right? We would create websites, we'd create apps, and the end user was human beings. And that's no longer the case. The agents, the AI agents are becoming the customer. So in today's episode, I'm going to give you a clear primer on what is changing on the internet and how you could actually make money, build products, and what you really need to know about this agentic era of the internet. I haven't seen anyone really post and do a de facto episode on it, so I figured I'd do it.

0:50So I tweeted this and I think this will resonate with you. I said, build startups for agents. Over the next 10 years, you're going to have a market of billions of customers, aka agents, with millions of wallets that want to use your services. Go look at every SaaS tool you use, Notion, Slack, Stripe, etc. Now ask, what is the version of this that's built purely for agents? Agent-native payments, agent-native communication, agent-native memory. Every single category gets rebuilt for agents. We're entering the machine-to-machine economy and almost nobody is building for it yet. So I tweeted this a few months ago and I started seeing more and more stuff come out.

1:40So let's actually go through what is happening, what is the shift, so that we really understand it. So the old web was humans searching, reading, comparing, clicking, and buying. And it was your website, your beautiful website, that was built to win the attention. But in the agent web, it's not for humans, right? It's agents discovering, evaluating, invoking tools, paying, renewing. So your company has to be machine usable. The way I think about it is the human customer wants persuasion. But the agent customer wants structured capability, permission, and trust. So how should we think about the agent buying journey?

2:29What does that even look like? I know it's weird even to say it, right? But it's important that we map it out so that we truly understand who the agent is, what are their desires, and what are the incentives that are going to get them to do things. Because if we figure that out, the people who figure that out are going to have companies that do a tremendous amount of revenue and value. And by the way, for people thinking, who are these agents? Well, everyone is going to have personal agents in their pocket, on the web, and they're also going to have business agents. You're starting to see this with Codex, Claude Code, OpenClaw, Hermes.

3:11I think it's pretty inevitable that agents are going to be everywhere and that's going to outnumber humans. So let's go back to the agent buying journey here. So for finding, find a payroll tool for 40 contractors. Then agents are going to basically do finding. Agents are going to be evaluating, like reading docs, pricing, APIs, reviews. Agents are going to be looking for trust. They're going to be checking policy. They're going to be checking limits. They're going to be asked their identity. Agents are going to be transacting. They're going to be paying. They're going to be booking. They're going to be signing.

3:50They're going to be subscribing. Agents are going to be using tools. They're going to be filing tickets. They're going to be changing settings. And agents are going to be recommending what other tools they should use. So they're going to be telling other agents what worked. And that's honestly the weirdest part about it all. We saw a glimpse of this with Moldbook. If you remember, Moldbook got acquired by Facebook. But it was a social network for agents. I think that was just the tip of the iceberg. You're going to see more and more social products that are specifically for agents. So the opportunity here is every step has missing infrastructure because the old internet assumed a person was doing the work or we're now in a new era.

4:38So what do agents need that humans do not? That's the obvious question to ask next, right? So they need identity. So who is this agent acting for? They need tools, right? What actions can it safely invoke? It needs an inbox. So where it could reply, OTPs, docs, and threadlands. Where does that all land? Memory. What does it know about my preferences and rules? Wallet. What can it spend and who approves it? So it's kind of similar, by the way, to an employee. As your employee gains trust, you might give them a credit card and then you might increase that limit and stuff like that. And then receipts, what did it see, decide, change, and buy?

5:31So let's actually talk through some really concrete examples because I think this will drive the point home to you. Have you seen Agent Mail? So Agent Mail is this customer support agent that has its own inbox, basically. Well, a better way to describe it, actually, is Agent Mail is inboxes for AI agents. So if you go to the website, it says, Agent Mail is the email inbox API for AI agents. It gives agents their own email inbox like Gmail does for humans. So they're basically building the agent infrastructure for email inboxes. Huge opportunity, backed by YC. I heard they're doing really well.

6:19Not surprised. Stripe recently launched basically a way to give your agent a wallet. And I think you're going to see more and more examples in fintech around this. So imagine a purchasing agent that can buy software with spend caps, approval rules, shared payment tokens, and an audit trail. Support. A user's agent files the ticket, includes logs, it asks for the refund, follows up, and escalates when ignored. So I think you're just going to see a lot more support agents and there's just a lot of infrastructure to be built around there. Procurement. A CFO agent compares 12 vendors, reads SOC 2 docs, negotiate terms, it recommends one that fits the policy.

7:12I'll give you an MCP server example. A SaaS app gives agents tools, searches customers, creates invoices, refund order, updates tickets, pulls reports. The agent doesn't scrape the UI. And one last one here, a local business example. A travel agent books dinner, changes the reservation, calls the hotel, pays the deposit, and updates the calendar. As I'm showing you these examples, like email inboxes for AI agents, and as you're starting to see, oh, I can understand an agent having a wallet, you might just visualize it more. For me, I need to see the examples to really, really understand it. And then that's when it really drives the point around, yeah, it makes sense that if agents are more agents than human beings on the internet, if agent traffic is more than human traffic on the internet, of course we need to create software, technology that services that, and there's a huge opportunity there.

8:17And also just our own websites and apps, and we're going to get into this next, is how do you basically create a homepage that is agent readable? So a human readable homepage on your website is you might have obviously a brand, a video, you'll have copy, you'll have social proof, demo, pricing. But an agent readable is you're going to have structured docs, schemas, you're going to have policies, examples, and endpoints. And agent readable means MCP tools, it means SDKs, it means OAuths, it means checkout, it means sandbox, it means receipts. And if an agent can't understand what you do and safely do something with you, you're basically invisible to them.

9:12So thinking about structuring your websites around agents and also creating a slash agents, your website slash agents and different entry points for agents is a huge opportunity. So what does this change for you? What does this change for people who are building startups, building businesses? One is that, and we've talked a lot about this on the channel, is that SEO is moving to AEO. So you want to optimize for agents deciding who to cite, trust, and recommend. So agents are of course going to go to other agents in the future asking, hey, what's a good website to do XYZ? but they're also just going to use with computer use things like chatGPT and things like perplexity and stuff like that.

10:10So you want to be top there. And if people are interested, I can do more in-depth how to actually rank on AEO. So subscribe and comment if you want me to do more there. Instead of having forums on your website, you want to do more tool calls. So the call to action becomes an action endpoint. Instead of having support docs, you want to actually do executable support. So the agents do the refund, they do the return, they do the reschedule, they troubleshoot, and they escalate. Instead of just having landing pages, you want to have capability manifests. So agents care less about slogans and more about what they can do with your product or service.

10:56Instead of just having sales calls, you want to have agent procurement. Buyers send agents to a short list before a human ever appears. Instead of just having analytics, you want to have agent analytics. Which agents visited, what did they ask, where did they fail, what did they bounce? Earlier in my career, I spent so much time in analytics and thinking about voice of customer. How do you get human voice of customer via survey data and pair that with things like Google Analytics to really, really increase your conversion rate from 1 % to 4 %? If you can do that, that would be the difference between millions of dollars and really just servicing the people coming to your website.

11:42But in a world where agents are coming to your websites, what does agent analytics look like? How do you optimize the conversion rate for agents? And that's just a really interesting question that I just can't stop thinking about. So this is the Startup Ideas podcast. I have to give you a few startup ideas just to get your creative juices flowing. There's a few startup ideas hiding in this shift. And I'll just give you rapid fire a couple or a few. Agent SEO agency, an agent identity and permissions technology, agent receipts and audit trails, agent-ready docs generators, agent inbox security, agent-readable pricing pages as a service, MCP servers for franchises, agent support desks, sandbox for agents to test SaaS.

12:33There's basically a bunch of ideas here that if you understand where the world is going in terms of agents, and I tried to explain it in the most clear possible way, you can start to envision a future internet where it's bifurcated into two, human internet and agent internet. And that's where we're going. And if you look at some of the most popular tools, some of the most popular apps and basically say, what's an agent version of that? there's a lot of opportunity over the next 10 years. That's my big prediction. Startups for agents. I want to see what you end up building. Hope this has been helpful.

13:19Hope this has been clear. Please let me know what you want to see on the next episode or on the next few episodes, who you want to see. I make these episodes for you to make it clear for you to understand where the world is going, what are ideas that can help you make money, be more productive, make impact. And if you notice why I didn't post a lot in the last couple weeks, I'd been sick. I'm starting to get back. My voice is getting there. And so there'll be more soon. Hope you have a creative day and I'll see you in the next one.