n8n Autonomous AI Restaurant Support & Order Agent
Autonomous AI agent workflow built in n8n leveraging Mistral Cloud LLM, Simple Memory session buffer, and live Google Sheets integrations for 24/7 menu lookups, inventory verification, and automated order booking.

Client Problem
MDA Restaurant experienced peak-hour customer support bottlenecks where manual staff spent hours handling repetitive chat inquiries regarding daily menu items, stock availability, and order intake instead of focusing on food preparation.
System Design
n8n visual workflow orchestrating an AI Agent node connected to Mistral Cloud Chat Model, Simple Memory session buffer, and dynamic Google Sheets tool endpoints (`Get Inventory`, `Get FAQ`, `Get Orders`) for real-time data reads and transactional appends.
Handling data accurately
- Real-time webhook trigger capturing incoming customer messages.
- Dynamic tool selection querying `Get Inventory` or `Get FAQ` based on intent classification.
- Transactional record append to `Get Orders` spreadsheet upon order confirmation.
Machine Learning & Agents
- AI Agent Node (ReAct Framework) — Dynamically parses incoming chat intents and invokes appropriate tools.
- Mistral Cloud Chat Model — Delivers high-precision natural language understanding and menu recommendations.
- Simple Memory Buffer — Maintains conversation state across multi-turn customer chat sessions.
- Google Sheets Tool Suite — Performs real-time stock checks (`read: sheet`) and logs customer bookings (`append: sheet`).
Conversation Flow
- Greeting & Options Presentation — Welcomes customer with interactive choices (Place order, View menu, Info, Order status).
- Live Inventory Query — Retrieves real-time stock counts (e.g. Veg Thali: 10, Chicken Thali: 15 available).
- FAQ & Policy Lookup — Answers inquiries regarding operating hours, delivery parameters, and dish details.
- Order Intake & Sheet Append — Collects items, confirms order, and writes record directly to live Google Sheet.
Measurable Outcomes & Cost Savings
Delivers sub-6-second (5.859s) responses using ~2,344 tokens per query. Eliminates 100% of routine menu inquiry overhead and provides 24/7 automated order booking.
Technologies Used
System in action
Future Improvements
- Integration with WhatsApp Business API and Telegram messaging webhooks.
- Stripe payment gateway integration for automated order pre-payment.
- PostgreSQL database migration for multi-branch concurrent order processing.
Ready to dive deeper?
Explore the architecture, commits, and implementation details of this project.