No Self-Serve "Pay Per Request" Path for AI Agents — Everyone Either Gets Stonewalled or Blocked, So Scraping and Absurd Workarounds Win

Summary

There is no simple, self-service way for an AI agent or a small developer to pay a site owner for the content it consumes. On the publisher side, the only commercial data path is an opaque, human-reviewed licensing process with no published criteria, no confirmation email, no per-request pricing, and a 1-3 week wait that usually ends in a silent 'no' — so developers who are perfectly willing to pay conclude they have no option but to scrape. On the agent side, the entire consumer web has responded by slamming the door shut: Delta kills automated booking bots, JPMorgan/Schwab lock down account access, Amazon shuts off shopping agents, and Cloudflare fingerprinting kills headless browsers, which has driven teams to rack up hundreds of physical iPhones with robotic styluses tapping capacitive glass just to order a ticket. Nothing in the middle exists: no HTTP-native, machine-readable price tag that an autonomous agent can discover, negotiate, and pay automatically per page/request. Site owners lose revenue and control, agents resort to evasion, and both sides burn enormous engineering effort on an arms race that a simple per-page price would eliminate. This is unsolved, extremely painful, and is exactly the gap a pay-per-crawl protocol fills — high willingness to pay on both sides.