Key facts
- A venture capitalist used an AI agent to try and secure a restaurant reservation on Resy.
- The AI agent's frequent requests led Resy to temporarily suspend the user's account.
- Resy stated that unapproved automated activity violates its terms of service and can pose risks.
- The user's account was reinstated with a warning against future violations.
JC Bahr-de Stefano, a principal at the fintech-focused venture capital firm Better Tomorrow Ventures, had his Resy account temporarily suspended after using an AI agent named Instinct to try and secure a reservation at the highly sought-after 4 Charles Prime Rib restaurant in New York City.
Instinct, an invite-only service, communicated with Bahr-de Stefano via text and was intended to check for open tables more frequently than a human could manually. However, the AI agent's activity resulted in approximately 200 API requests per hour to Resy's website, including checking availability every 10 minutes and every 0.4 seconds around the time reservations typically drop.
An American Express spokesperson, which owns Resy, stated that the platform does not permit unapproved third-party bots or agents, as such automated activity can introduce risks and compromise a fair reservation experience. Resy allows users to find restaurants and make reservations through integrations with OpenAI's ChatGPT and Anthropic's Claude.
Bahr-de Stefano noticed the issue on Sunday when he could not access his account and received an email from Resy stating his account displayed behavior in violation of its Terms of Service. After reviewing Instinct's activity log, he understood the account was flagged for spam-like behavior. His account was reinstated on Tuesday with a warning that further violations could lead to permanent closure of his Resy account and American Express credit card.
Bahr-de Stefano indicated he plans to continue using Instinct but will implement parameters to control its website ping frequency and will request the AI to outline its plan before execution. He also suggested that platforms like Resy need to differentiate between legitimate AI agents acting on behalf of users and malicious bots.
Discussion