Resy cuts off user whose AI assistant made 200 requests an hour
Reports of deactivations across Resy, Amex and Uber suggest platforms are starting to police automated agents that hit their systems at machine speed.
Resy deactivated a user after an automated assistant acting on his behalf made 200 requests an hour, according to a report by Inc.
The same report says users of Resy, Amex and Uber have described a growing crackdown on AI agents, and that a single agent appears to connect all three cases.
The agent has not been named publicly, and the platforms have not published the thresholds at which they act.
For founders building agentic products, the mechanics matter more than the outrage. An agent that polls a restaurant booking system a few times a second looks, from the platform's side, like abuse rather than a customer.
The account that gets deactivated is usually the user's, not the developer's. That is the part worth reading twice: the cost of an aggressive polling loop lands on the person who installed the software.
Terms of service on consumer platforms generally predate agents, and the platforms are under no obligation to rewrite them before enforcing them. A product whose core function depends on automating a third party's logged-in experience is a product with a dependency it does not control.
The practical questions are the dull ones. How many requests an hour does the agent make at peak, does it back off when a platform slows it down, and what does the user see when an account is cut off.
Resy, Amex and Uber have not set out how many accounts have been affected.