The AI Agent Fiasco: From Restaurant Ban to Reinstatement




Navigating the Digital Dining Landscape: The Perils and Promise of AI Assistants
Embracing Emerging Technologies: A Personal Journey with AI
Driven by an innate passion for technological innovation, I constantly experiment with the latest advancements. My journey began with early access to ChatGPT and continued through exploration of advanced AI models like Claude and Codex. My prior professional endeavors centered on the evolution of intelligent systems, shaping my deep engagement with artificial intelligence.
The Evolution of My AI Ecosystem: From OpenClaw to Hannah
My initial foray into personal AI agents involved setting up an OpenClaw instance on a Mac Mini to streamline daily tasks. Despite its complexity in the early stages, it laid the groundwork for my current, more sophisticated system. I transitioned to Hermes, also hosted on a Mac Mini, and over several months, cultivated an extensive network of AI agents. This ecosystem is overseen by a primary agent named Hannah, a tribute to my long-serving executive assistant, through which I interact via Telegram. Hannah orchestrates various specialized sub-agents, including one dedicated to tracking sports, which proved invaluable during the Knicks' recent impressive season, and another, dubbed 'Concierge,' focused on managing events, restaurants, and reservations.
The Unexpected Repercussions: An AI Agent's Automated Actions Lead to a Resy Ban
My wife and I harbored a desire to revisit the acclaimed New York City establishment, 4 Charles. Confident in my AI's capabilities, I entrusted an agent with my Resy account credentials, instructing it to secure a reservation as soon as availability emerged. Unbeknownst to me, I had failed to implement any restrictions on the agent's activity. Its continuous, unchecked requests to the platform were quickly identified as bot traffic. Within two days, my Resy account was deactivated. This incident underscored the critical role a Resy account plays in New York City's dining scene, a realization that only truly hits once access is lost.
Exploring Alternatives: The High Cost of a Banned Account
Following the ban, I was compelled to subscribe to Dorsia's premium tier, an alternative reservation service for high-end restaurants, which required a significant financial commitment. While it offered a temporary solution, the inventory of available restaurants was notably more limited. My wife, though understanding of my technological explorations, was less than thrilled by the inconvenience, and we occasionally resorted to using her Resy account.
Initial Attempts at Reinstatement: Unresponsive Communications
Immediately after the account deactivation, I contacted Resy, expressing my apologies and attempting to explain the situation. My initial emails, numbering four or five, either went unanswered or received terse, one-sentence replies. Rumors circulated that accounts were reinstated after a month, so I patiently waited, but the deadline passed without resolution.
A Strategic Appeal: Leveraging AI for Account Restoration
In a moment of ironic inspiration, I tasked my primary AI agent, Hannah, with composing an appeal to Resy. Hannah generated a meticulously crafted, multi-paragraph email that was professional and detailed. It outlined my actions, acknowledged my mistake, and assured compliance with future policies. The email also referenced my long-standing history as a Resy customer and my status as an American Express Business Platinum cardholder. This professionally worded communication proved effective; I received a response within two days, and my account was reinstated.
Lessons Learned: The Imperative of Guardrails for AI Agents
The experience was both exciting and a profound lesson. I now recognize the critical importance of establishing robust guardrails for AI agents to prevent unintended consequences. I am now far more diligent about ensuring my agents interact with external platforms responsibly and in accordance with best practices. Furthermore, I consult with my engineering team to avoid similar missteps. As AI agents become more prevalent, understanding and respecting the terms of service of various online platforms is paramount, given that many companies are still formulating their strategies for managing AI-driven interactions.