A video shared by tech enthusiast Tech by Pawan has prompted questions about how far an AI agent can go while using a public booking website. According to a Gulte report published on September 29, Pawan asked the agent to seek a Tirumala room for December 21. The video reportedly shows it signing in with details supplied by the user, navigating the TTD website and selecting options before reaching the payment stage.

The demonstration did not result in a confirmed reservation. Pawan said the process stopped because bank or payment details were not provided. A TTD IT official also reportedly said that accommodation cannot be booked through the website without a valid darshan ticket. Those two details matter: reaching a payment page is different from satisfying every booking requirement and receiving a room confirmation.

The report calls the tool “ChatGPT-6 Astra,” but the video and article do not independently establish that product name. OpenAI’s published documentation describes ChatGPT agent as capable of navigating websites and filling forms on a user’s behalf, with safeguards that include user confirmation for high-impact actions.

What the Tirumala booking video shows

Based on the account provided, the agent acted after a person instructed it to look for accommodation and supplied login details. It moved through steps a user would normally complete on the website. The reported attempt ended before payment, and there is no evidence in the account that a room was booked or that the agent bypassed TTD’s eligibility rules.

That makes the video a demonstration of automated website navigation, rather than proof of unauthorised access. The broader claim in the original report about AI bots bypassing government security controls is not established by this booking example. The account provides no technical evidence of a security breach on the TTD site.

Why the distinction matters

AI agents can make online tasks faster by searching pages and entering information under a user’s direction. Booking systems still have their own requirements, including eligibility checks, payment and final confirmation. OpenAI says its agent includes confirmation and supervision measures for consequential actions, while acknowledging that users should monitor what it does.

The Tirumala video raises a practical question about how booking websites and users manage automated assistance. On the facts reported so far, the clearest outcome is limited: an AI agent reportedly navigated much of a room-booking flow, then stopped without completing payment or securing accommodation.