Analysis
Cloudflare's Pay-Per-Crawl: 3 Businesses You Can Build Now
Cloudflare just built payment rails for AI agents, and Greg Isenberg lays out three businesses you can start building before the space gets crowded.
The Old Web Is Breaking, and That's the Opportunity
For two decades, the internet ran on a simple trade: search engines crawled your site, sent you visitors, and you monetized that attention with ads, emails, or subscriptions. AI agents break that trade. An agent can read your page, extract the answer, and hand it to a user who never visits your site — so the content still creates value, but the visit, and the monetization that came with it, disappears.
Cloudflare's response is a set of tools that let site owners see, control, and now charge AI crawlers for access. The pieces include AI Crawl Control (visibility into crawler activity), Pay Per Crawl (site owners can charge per access), and a broader Monetization Gateway that extends payment rules to APIs, datasets, and tool calls — not just web pages.
- Understand the shift: agents monetize useful resources, not attention, so the business model has to change with them.
- Watch the payment rail: it uses HTTP 402 ("Payment Required") responses, verified at Cloudflare's edge, so a request can become a transaction without checkout flows or sales calls.
Three Businesses to Build on Top of It
The underlying idea across all three: agents need clean, trusted, structured data and expertise to do their jobs well, and someone has to build the pipes that deliver it.
The first idea is a niche data refinery — pick one niche where valuable information is messy and scattered (competitor pricing, reviews, hiring signals, permits), collect it manually into a spreadsheet for a small set of businesses, then package it into reports. The early customer isn't the business owner directly — it's the agencies and consultants who already sell into that niche and can charge more with better data.
The second is agent readiness for businesses — essentially SEO for AI agents. Run buyer-intent prompts across AI tools to see how a company shows up (wrong pricing, missing docs, bad comparisons), show the founder the screenshot, then sell the fix: clean documentation, structured FAQs, honest comparison pages, an LLM-readable file. Start as a paid audit, then productize once patterns repeat across clients.