Restaurant Table Booking Agent

Our project is a fully autonomous, voice-enabled restaurant booking agent that handles real-world reservation complexity—multi-table seating, large groups, private-room deposits, and even full-venue buyouts—using natural conversation alone. The user simply speaks, and the agent interprets intent, checks availability, computes table combinations, triggers Stripe deposits when needed, and finalizes bookings end-to-end.
The system replaces manual front-of-house workload by automating call handling, capacity reasoning, and payment collection. In our demo, the ElevenLabs voice agent gathers reservation details, uses backend FastAPI tools (find_tables, create_reservation) for availability, triggers an n8n workflow to generate live Stripe Checkout links for deposits, and completes the reservation once payment is confirmed.
This project stands out because it integrates real conversational autonomy with multiple live systems: ElevenLabs Agent Mode, a capacity-aware booking engine, ngrok-exposed APIs, n8n automation, and Stripe Checkout. It reliably supports normal bookings, multi-table groups, private-room requests, and full-restaurant buyouts.
Tech Stack:
ElevenLabs Agent Mode, Python/FastAPI backend, ngrok, n8n workflows, Stripe Checkout API, and a custom restaurant table dataset with capacity attributes.
Reproducibility:
Run the FastAPI server, expose it via ngrok, connect tools in ElevenLabs, link the n8n Stripe workflow, and test bookings—standard, multi-table, private-room (deposit flow), or full buyout. The entire experience works live through natural voice interaction.