https://github.com/mxcaoalina/datemeal_mobile
DateMeal is a mobile AI assistant that helps users discover ideal date restaurants via natural chat.
https://github.com/mxcaoalina/datemeal_mobile
ai-agent fastapi react-native
Last synced: 4 months ago
JSON representation
DateMeal is a mobile AI assistant that helps users discover ideal date restaurants via natural chat.
- Host: GitHub
- URL: https://github.com/mxcaoalina/datemeal_mobile
- Owner: mxcaoalina
- Created: 2025-04-22T21:47:56.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2025-04-30T20:42:21.000Z (about 1 year ago)
- Last Synced: 2025-04-30T20:43:08.681Z (about 1 year ago)
- Topics: ai-agent, fastapi, react-native
- Language: Python
- Homepage:
- Size: 25.6 MB
- Stars: 2
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# DateMeal Mobile App - Full Stack Documentation
---
## Project Overview
**DateMeal** is an AI-powered mobile application that helps users find ideal restaurants for a date by chatting with a personalized recommendation agent. The app tailors suggestions based on mood, ambience, budget, cuisine preferences, and dietary needs.
With a delightful onboarding experience, natural-language chat, and a sleek mobile UI, DateMeal guides users from selection to detail view—and even saves favorites. It uses Azure OpenAI (GPT-4o) to generate recommendations and optionally enriches them with real-world data via Bing Web Search Grounding.
---
## Why It's Innovative
- Uses **conversational AI** (Azure OpenAI) to recommend restaurants, instead of forms or filters.
- Supports **real-time refinement** via follow-up chat.
- Applies **Human-in-the-Loop design** for interactive, evolving suggestions.
- Augments AI responses with **Bing Web Search** for real-world accuracy and visuals.
---
## Responsible AI
- All AI-generated content is reviewed by the user before action—ensuring human oversight (Human-in-the-Loop).
- User preferences and past interactions are stored locally using AsyncStorage to preserve privacy.
- GPT-4o prompts are explicitly scoped to avoid biased, harmful, or sensitive content.
- The app avoids storing personal data in the cloud and includes fallback logic to ensure safety in the event of AI errors.
---
## Real-World Use Case
DateMeal solves a familiar but often frustrating problem: *“Where should we eat tonight?”*
By offering personalized suggestions based on context (budget, mood, vibe, etc.) via natural conversation, it removes friction from decision-making. We envision this tool being used by:
- Couples planning a date night
- Groups deciding on dinner while traveling
- Users with dietary restrictions looking for safe dining options
The mobile-first, chat-based experience fits seamlessly into users’ lives—like having a restaurant-savvy friend in your pocket.
---
## Tech Stack
| Layer | Technology |
|----------------|----------------------------------------------|
| Mobile Frontend | React Native (Expo), TypeScript, Zustand |
| Backend API | FastAPI (Python), Azure OpenAI, Bing Search |
| Storage | AsyncStorage (local) |
| Hosting | Azure App Service (Backend) + EAS (Frontend) |
| AI Services | Azure OpenAI (GPT-4o), Bing Web Search |
| Styling | Custom theming + StyleSheet |
---
## System Architecture
```
User (Mobile App)
↓
React Native (Expo)
↓
FastAPI Backend (Azure App Service)
↓
Azure OpenAI (GPT-4o)
⇳ (optional) Bing Web Search Grounding
```
- The mobile client calls the FastAPI backend.
- The backend constructs a structured prompt, sends it to Azure OpenAI, and parses the structured JSON response.
- If available, the backend enriches results with real images and links via Bing Search API.
---
## Project Structure
### Frontend (`app/`)
```plaintext
app/
├── components/ # Reusable UI elements
├── navigation/ # Stack navigator
├── screens/ # Chat, Onboarding, Details, Home
├── services/ # API adapters (axios)
├── store/ # Zustand + AsyncStorage
├── theme/ # Style constants
├── types/ # Shared interfaces
├── utils/ # API base URL config, helpers
├── App.tsx # Entry point
```
### Backend (`backend-python/`)
```plaintext
backend-python/
├── api/ # API route handlers
│ ├── advise.py
│ ├── health.py
│ └── refine.py
├── models/ # Pydantic schemas
├── services/ # Azure OpenAI & Bing search wrappers
├── utils/ # Logger setup, etc.
├── main.py # FastAPI app
├── requirements.txt # Dependencies
├── .env # API keys and configs
```
---
## System Flow Overview
```
+--------------------+ +--------------------+
| React Native App | | AsyncStorage |
|--------------------| |--------------------|
| - User Onboarding |<------->| - Preferences |
| - Chat Interface | | - Cached Results |
+--------+-----------+ +---------^----------+
| |
| HTTP Request |
v |
+--------+-----------+ +---------+----------+
| FastAPI | | Azure OpenAI (GPT) |
|--------------------| |--------------------|
| - Receives input | | - Chat Completion |
| - Formats prompt |-------->| - Returns JSON |
| - Handles fallback | +--------------------+
+--------+-----------+
|
v
+--------------------+ +--------------------+
| Bing Search API |-------->| Restaurant Data |
|--------------------| | (Structured JSON) |
| - Image lookup | +--------------------+
| - URL enrichment |
+--------------------+
```
---
## API Endpoints
| Method | Endpoint | Purpose |
|--------|----------------------|--------------------------------------------|
| GET | `/health` | Health check |
| POST | `/advise` | Generate restaurant recommendations |
| POST | `/restaurant/refine` | Refine a previous recommendation |
---
## Key Frontend Features
- **Onboarding Flow**: Set preferences (mood, party size, cuisine, vibe, budget, allergies).
- **Chat Interface**: Converse with DateMeal AI Agent to discover ideal restaurants.
- **Restaurant Detail View**: Menus, hours, contact, map links.
- **Save for Later**: Saved restaurants displayed on home.
- **Offline Storage**: Preferences and past results stored via AsyncStorage.
---
## Key Backend Features
- **AI-Powered JSON Recommendations**: GPT-4o returns structured JSON.
- **Bing Enrichment (optional)**: Uses Bing Search API to add live URLs and photos.
- **Validation**: Strong schema enforcement with Pydantic.
- **Graceful Fallbacks**: If GPT fails, user still receives a static response.
- **CORS**: Enabled for Expo device access.
---
## Prompting Strategy
The backend sends a structured system prompt with examples, instructing GPT-4o to return exactly 3 restaurant recommendations in valid JSON format with specific fields such as:
- `name`, `description`, `cuisine`, `priceRange`, `rating`
- `menuItems` (array of dishes)
- `highlights`, `website`, `location`
> **Example System Prompt**:
>
> You are a restaurant recommender AI. Based on the following user preferences, return 3 diverse restaurant suggestions in strict JSON format.
> - Cuisine: Korean
> - Budget: $$
> - Mood: Cozy
It uses dynamic input from the user (e.g., "Romantic Italian in Brooklyn") to populate the prompt. If the response is malformed or missing, fallback handling is triggered.
---
## App Screens & Demo Walkthrough
**▶ [Watch the 3-min DateMeal demo](https://drive.google.com/file/d/1_zzxjOtndYJLJsYm19UYlWKZYLyoMRw1/view?usp=share_link)** – user flow + architecture snapshot
Home Screen: Displays a greeting, shows saved restaurants and quick actions like chatting or editing preferences.
Preferences Page: Update dietary restrictions, must-not-haves, and more to refine future results.
Onboarding Flow: Users specify vibe, cuisine, party size, budget, dietary needs, and no-gos before receiving tailored results.
Chat Assistant: Conversational UI lets users ask for suggestions and receive real-time responses from the AI agent.
Refine Recommendations: After receiving suggestions, users can refine results by asking the agent to adjust based on new criteria.
Restaurant Detail View: Expanded view of a selected restaurant, including highlights, contact info, hours, and curated menu.
---
## Deployment
- **Backend**: Deployed to Azure App Service — [View Deployment](https://datemeal-backend-cga2d8fqfsctesh9.eastus2-01.azurewebsites.net/)
- **Frontend**: Published via Expo EAS Update — [Expo Dashboard](https://expo.dev/accounts/mxcao_alina/projects/DateMealMobile)
Scan to preview the latest DateMeal app
---
## Running Locally
### 1. Start Backend
```bash
cd backend-python
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
uvicorn main:app --reload --port 8001
```
### 2. Configure Frontend (for device testing)
Edit `utils/networkUtils.ts`:
```ts
const YOUR_COMPUTER_IP = '192.168.x.x';
```
### 3. Run Expo Frontend
```bash
npm install
npx expo start
```
Scan QR code to run on mobile or simulator.
---
© AI Agent Hackathon 2025 · Team: mxcaoalina, ziqun-liu, RajasekharwhY