AI Services Market › Use cases › Travel & Bookings
Travel & BookingsDone with custom-built or unnamed agent Mar 18, 2026
Book restaurant reservations end to end on real sites
The poster built agents that find a restaurant, fill out the reservation form and confirm the booking on a real site, handling dynamic elements and bot detection. · Mar 18, 2026
- What They Did
- The poster says their agents completed restaurant booking on the live site rather than a scripted demo: the agent found a restaurant, filled the reservation form, and confirmed it. They describe the setup as a stealth browser with persistent profiles and fingerprinting, residential proxies, and CAPTCHA-solving; they also say it still needed manual setup and was not yet a packaged product.
- How It Went
- They report the flow worked end to end on real sites and could get through dynamic elements and bot detection. A stated limitation was payments: their restaurant tests avoided it because booking did not require payment, and Uber used a saved card.
- Worth Knowing
- The main catch is that the current stack was still manual, with separate pieces duct-taped together rather than something an OpenClaw user could just install and run.
Try it yourself with custom-built or unnamed agent
Find a [cuisine] restaurant in [city] with a table for [party size] on [date] around [time], rated at least [rating], and show me your top three options. After I pick one, go through its reservation page and fill in the form with my name [name], phone [phone] and email [email]. Pause before submitting and show me the exact details, then once I approve, submit and give me the confirmation number or screenshot.