We are currently working on SelfDine AI, a restaurant POS platform with an integrated AI-powered guest ordering chatbot.
The goal is to connect the customer ordering experience with the restaurant's daily operations—from menu discovery and ordering to kitchen coordination, payments and reporting.
AI-Powered Guest Ordering
The guest chatbot allows customers to search the menu using natural language, get recommendations, manage their cart and complete checkout.
It also supports customer preferences such as cuisine, dietary restrictions, spice level, protein choice, flavor profile and allergies.
How the AI Search Works
Instead of relying on one search method, the chatbot uses a tiered approach:
Intent classification
Fuzzy search
Vector search
Semantic / AI profile search
LLM fallback
The recommendation system can also consider factors such as budget, customer profile, weather, meal time, kitchen load and conversion probability.
Connecting AI With Restaurant Operations
SelfDine AI also connects the guest experience with restaurant operations.
A customer can scan a table QR code and enter an AI ordering session. After the order is placed, it can move through staff confirmation and kitchen preparation.
The platform also includes POS, menu management, inventory, payments, table management, kitchen and waiter operations, staff management, dashboards and reports.
Current Project Scale
The documented codebase currently includes around 61 controllers, 61 models and 165 service classes. The frontend contains approximately 211 pages across different modules.
The technology stack includes Laravel 11, Vue 3, MySQL, Laravel Sanctum, and AI/LLM technologies including OpenAI, Ollama and Groq.
We are continuing to work on SelfDine AI and improve how AI can fit into practical restaurant workflows.
The focus is simple: make restaurant ordering more intelligent, personalized and connected.
Top comments (0)