Clearest Explanation of AI Agents as Customers
The speaker argues that AI agents are becoming the primary 'customers' of the internet, replacing human users as the dominant force driving web traffic and commerce. This shift requires entirely new infrastructure — from agent-native inboxes and wallets to machine-readable websites — creating massive startup opportunities. The speaker frames this as a bifurcation of the internet into a human layer and an agent layer.
Summary
The speaker opens by identifying a major underreported shift: AI agents are becoming the primary users of the internet, displacing humans as the default 'customer.' Where the old web was built around human attention — beautiful websites, persuasive copy, social proof — the emerging agent web requires structured capability, permissions, and trust instead of persuasion.
The speaker maps out the 'agent buying journey,' describing how agents will handle the full commercial lifecycle: finding services, evaluating options, verifying trust and policy compliance, transacting (paying, booking, signing), using tools, and even recommending services to other agents. This last step — agents advising other agents — is described as the 'weirdest' but most significant emerging dynamic, with a reference to Moltbook (a short-lived social network for agents acquired by Facebook) as an early glimpse of this trend.
The speaker then outlines what agents need that humans do not: identity (who the agent acts for), tools (what actions it can safely invoke), an inbox (where communications land), memory (preferences and rules), a wallet (spend limits and approvals), and receipts (audit trails of decisions and purchases). This is compared to the trust relationship between an employer and employee, where expanded trust leads to expanded privileges like a corporate credit card.
Concrete examples are provided to ground the concepts: AgentMail (a YC-backed startup providing email inboxes for AI agents), Stripe's agent wallet product, a CFO agent comparing vendors and reading SOC 2 docs, and a travel agent booking and managing reservations autonomously. These examples illustrate where agent infrastructure is already emerging and where gaps remain.
The speaker then discusses how businesses should adapt: moving from human-readable homepages to agent-readable ones with structured docs, schemas, policies, and API endpoints; replacing forms with tool calls; replacing support docs with executable support agents; and replacing landing pages with 'capability manifests.' SEO is described as evolving into AEO (Agent Engine Optimization), where the goal is to be trusted and recommended by agents.
The episode closes with a rapid-fire list of startup ideas born from this shift, including agent SEO agencies, agent identity and permissions technology, agent receipt and audit trail tools, agent-readable pricing page generators, MCP servers for franchises, and sandboxes for agents to test SaaS products. The speaker predicts a bifurcated internet — human and agent — and calls the next 10 years a major opportunity window for founders who build for agents.
Key Insights
- The speaker argues that the agent customer fundamentally differs from the human customer in motivation: humans want persuasion, while agents want structured capability, permission, and trust — meaning traditional marketing and UX design are largely irrelevant for agent-facing products.
- The speaker claims agents will not only use services but will recommend services to other agents, creating a machine-to-machine word-of-mouth dynamic — pointing to Moltbook, a social network for agents acquired by Facebook, as an early prototype of this behavior.
- The speaker identifies AgentMail — a YC-backed startup providing dedicated email inboxes via API for AI agents — as concrete early evidence that agent-native infrastructure is already being built and gaining traction.
- The speaker argues that if a company's website cannot be understood or safely used by an agent, that company is effectively invisible to agent traffic — framing agent-readability (structured docs, schemas, MCP tools, OAuth, sandboxes) as the new baseline for online discoverability.
- The speaker predicts that SEO is transitioning to AEO (Agent Engine Optimization), where the goal shifts from ranking in search engines for human queries to being trusted, cited, and recommended by AI agents — and expresses interest in covering this topic in a dedicated future episode.
Topics
Transcript
[0:00] There's a really big shift happening right now on the internet and I don't think people are talking about it. You know, for the longest time, the user of the internet were human beings. They were us, right? We would create websites, we 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 [0:30] 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…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Greg Isenberg
What are Agentic Loops?
Professor Ross Mike explains agentic loops — AI systems that self-iterate without human input — and argues they are largely impractical for most developers due to high token costs and poor assumption-making. He distinguishes between 'human in the loop' (iterative, guided development) and fully automated loops, recommending loops only for constrained, binary-feedback processes like code review. He shares his own GP loop use case as a rare valid example.
How to use Obsidian with Claude in 61 seconds
This short tutorial outlines three levels of using Obsidian with Claude AI to accelerate professional learning and career resilience. It covers connecting Obsidian to Claude via MCP, using the Smart Connections plugin to identify skill gaps, and prompting Claude to generate a public learning plan based on your notes.
Been playing with this cool ai tool from genspark
The video demonstrates how GenSpark's AI tool can automate LinkedIn job post lead generation by identifying decision makers and drafting personalized cold emails. It then presents a business model called 'Outcome as a Service,' where users build competitor-tracking workflows and sell daily briefings to clients for recurring revenue.
Hermes Agent Explained
The transcript introduces Hermes, an AI agent that retains memory across sessions unlike standard LLMs. It highlights key features including cost-efficient reasoning via Open Router with Quen 3.6 Plus, Obsidian integration for persistent context, and cron job automation for token-free repetitive tasks. The entire stack is noted to run on an Android phone.
How to make really good Claude skills (clearly explained in 42 seconds)
The transcript outlines a method for creating high-quality Claude skills by providing context, defining clear criteria, iterating until successful, and letting the AI write the skill itself. The process emphasizes testing and collaborative debugging to ensure reliability. The speaker argues AI-written skills outperform human-written ones because they reflect what actually worked.