Welcome to the AI-Powered Travel Agent Automation project! This project is a Streamlit application designed to automate the process of travel agent tasks, including managing travel enquiries, generating itinerary suggestions, inputting vendor replies, and generating professional-looking quotations in PDF and DOCX formats. The system leverages Large Language Models (LLMs) for content generation and Supabase for backend data storage and file management.
This project is a AI Travel Agent demonstrating an AI-powered travel automation system. It allows users to manage travel enquiries, generate itinerary suggestions, input vendor replies, and automatically create professional-looking quotations in PDF and DOCX formats. The system leverages Large Language Models (LLMs) for content generation and Supabase for backend data storage and file management.
- Enquiry Management:
- Submit new travel enquiries with details like destination, duration, number of travelers, trip type, and client information.
- AI Itinerary Suggestions:
- Generate AI-powered suggestions for places and attractions based on enquiry details.
- Selectable AI providers (Google Gemini, OpenRouter, Groq, Together.AI).
- Vendor Reply Integration:
- Input and store vendor replies, including pricing, inclusions, and exclusions.
- AI Quotation Generation:
- Automatically generate structured quotation data using LLMs based on enquiry, itinerary, and vendor reply.
- Produce downloadable PDF quotations with a professional layout.
- Convert generated PDFs to DOCX format.
- Cloud Storage:
- Store generated quotation documents (PDF, DOCX) in Supabase Storage.
- Data Persistence:
- All enquiries, client details, itineraries, vendor replies, and quotation metadata are stored in a Supabase PostgreSQL database.
- Configurable AI:
- Choose between different AI providers (Gemini, OpenRouter, Groq, Together.AI) for content generation tasks.
- Specify default models for OpenRouter, Groq, and Together.AI via environment variables.
- Frontend: Streamlit
- Backend & Database: Supabase (PostgreSQL, Storage)
- AI/LLM Integration:
- Langchain
- LangGraph (for orchestrating quotation generation steps)
- LLM Providers: Google Gemini, OpenRouter, Groq, Together.AI
- Document Generation:
- FPDF2 (
fpdf) for PDF creation pdf2docxfor PDF to DOCX conversion
- FPDF2 (
- Environment Management:
python-dotenv - Programming Language: Python 3.11
.
├── app.py # Main Streamlit application entry point
├── requirements.txt # Python dependencies
├── .env.example # Environment variable template
├── .python-version # Python version specification
├── schema.sql # Database schema
├── schema-drop.sql # Database schema drop script
├── storage.sql # Storage configuration
│
├── assets/ # Static assets
│ ├── fonts/ # Custom fonts for PDF generation
│ │ ├── DejaVuSansCondensed.ttf
│ │ ├── DejaVuSansCondensed-Bold.ttf
│ │ └── DejaVuSansCondensed-Oblique.ttf
│ ├── top_banner.png # Banner image for PDF header
│ └── tripexplore-logo-with-rating.png # Logo for PDF header
│
├── src/ # Source code package
│ ├── __init__.py # Package initialization
│ ├── models.py # Data models
│ │
│ ├── core/ # Core business logic
│ │ ├── __init__.py
│ │ ├── itinerary_generator.py # AI-powered itinerary generation
│ │ └── quotation_graph_builder.py # LangGraph workflow for quotations
│ │
│ ├── llm/ # LLM related functionality
│ │ ├── __init__.py
│ │ ├── llm_prompts.py # Prompt templates and configurations
│ │ └── llm_providers.py # LLM provider integration (Gemini, OpenRouter, Groq)
│ │
│ ├── ui/ # User interface components
│ │ ├── __init__.py
│ │ ├── sidebar.py # Global sidebar configuration
│ │ ├── ui_helpers.py # Shared UI helper functions
│ │ │
│ │ ├── components/ # Reusable UI components
│ │ │ ├── __init__.py
│ │ │ ├── tab3_actions.py # Action handlers for tab3
│ │ │ └── tab3_ui_components.py # UI components for tab3
│ │ │
│ │ └── tabs/ # Tab implementations
│ │ ├── __init__.py
│ │ ├── tab1_new_enquiry.py # New travel enquiries tab
│ │ ├── tab2_manage_itinerary.py # Itinerary management tab
│ │ └── tab3_vendor_quotation.py # Vendor quotation tab
│ │
│ └── utils/ # Utility functions
│ ├── __init__.py
│ ├── constants.py # Application constants
│ ├── docx_utils.py # PDF to DOCX conversion
│ ├── pdf_utils.py # PDF generation logic
│ └── supabase_utils.py # Supabase integration utilities
.
- Python 3.11
- A Supabase account (for database and storage)
- API Keys:
- Google API Key (if using Gemini)
- OpenRouter API Key (if using OpenRouter)
- Groq API Key (if using Groq)
- Together.AI API Key (if using Together.AI)
- Clone the Repository:
git clone <repository-url> cd <repository-directory>
- Create a Virtual Environment and Activate it:
python -m venv venv # On Windows venv\Scripts\activate # On macOS/Linux source venv/bin/activate
- Install Dependencies:
pip install -r requirements.txt
- Set up Environment Variables:
- Copy
.env.exampleto a new file named.env:cp .env.example .env
- Edit
.envand fill in your actual credentials:SUPABASE_URL="YOUR_SUPABASE_URL" SUPABASE_KEY="YOUR_SUPABASE_ANON_KEY" # Or service_role key if needed for specific policies GOOGLE_API_KEY="YOUR_GOOGLE_API_KEY" 9A67 OPENROUTER_API_KEY="YOUR_OPENROUTER_API_KEY" OPENROUTER_DEFAULT_MODEL="google/gemini-flash-1.5" # Example, can be overridden OPENROUTER_HTTP_REFERER="http://localhost:3000" # Optional, set your app's URL OPENROUTER_APP_TITLE="AI Travel Quotation" # Optional, set your app's title GROQ_API_KEY="YOUR_GROQ_API_KEY" GROQ_DEFAULT_MODEL="llama3-8b-8192" # Optional, specify default Groq model TOGETHERAI_API_KEY="YOUR_TOGETHERAI_API_KEY" TOGETHERAI_DEFAULT_MODEL="meta-llama/Llama-3.3-70B-Instruct-Turbo-Free" # Optional, supported: "meta-llama/Llama-3.3-70B-Instruct-Turbo-Free", "deepseek-ai/DeepSeek-R1-Distill-Llama-70B-free"
- Copy
- Supabase Setup:
- Go to your Supabase project dashboard.
- Database:
- Navigate to the "SQL Editor".
- Click "+ New query" and run the contents of
schema.sqlto create the necessary tables. - (For development/reset, you can use
schema-drop.sqlto remove tables.)
- Storage:
- Navigate to "Storage".
- Click "Create a new bucket".
- Name the bucket
quotations. Make it a public bucket if you want direct public URLs, or keep it private and rely on signed URLs (the app handles both). - Go back to the "SQL Editor" and run the contents of
storage.sqlto set up access policies for thequotationsbucket. This allows anonymous uploads and reads as configured in the file.
- Prepare Assets (for PDF Generation):
- The PDF generation uses custom fonts and images.
- The DejaVu fonts are included in
assets/fonts/.
- You need to provide the following image files in the
assets/directory for full visual functionality:assets/top_banner.png(a banner image for the top of the PDF)assets/tripexplore-logo-with-rating.png(a logo image)
- These image files are optional - if not present, the PDF generation will print warnings and skip them, but all core functionality will still work.
Once the setup is complete, run the Streamlit application:
streamlit run app.pyThe application will open in your web browser, typically at http://localhost:8501.
The application is organized into three main tabs:
- 📝 New Enquiry:
- Fill in the travel details (Destination, Number of Days, Travelers, Trip Type).
- Provide client information (Name, Mobile, City, Email).
- Click "Submit Enquiry" to save the information to Supabase.
- 🔍 Manage Enquiries & Itinerary:
- Select an existing enquiry from the dropdown list.
- View the details of the selected enquiry.
- If an AI-generated itinerary already exists, it will be displayed.
- Click "Generate Places Suggestions" to use an LLM (selected via the sidebar AI Configuration) to suggest places/attractions for the trip. The suggestions will be saved.
- ✍️ Add Vendor Reply & Generate Quotation:
- Select an enquiry. Its details and any existing itinerary/vendor reply will be loaded.
- Add/View Vendor Reply:
- Input or update the vendor's reply, including details like pricing, inclusions, exclusions, hotel information, etc.
- Click "Submit/Update Vendor Reply" to save it.
- AI Quotation Generation:
- Ensure a vendor reply is available.
- The system uses the enquiry details, AI-generated itinerary (from Tab 2 or database), and the vendor reply to generate a quotation.
- Click "Generate Quotation PDF" to:
- Use an LLM to structure all information into a JSON.
- Generate a PDF document from this JSON.
- Upload the PDF to Supabase Storage.
- Save the quotation metadata (including the JSON and storage path) to the database.
- Click "Generate Quotation DOCX" to:
- Perform the same LLM structuring and PDF generation as above.
- Convert the generated PDF bytes into DOCX format.
- Upload the DOCX to Supabase Storage.
- Update the quotation record in the database with the DOCX storage path.
- Download/View Quotation Files:
- Links to download/view stored PDF and DOCX files from Supabase Storage will appear (public or signed URLs).
- Buttons to download locally generated PDF/DOCX files (from the current session before potential re-runs or selection changes) will also be available.
- Use the sidebar to select the AI Provider (Gemini, OpenRouter, Groq, or Together.AI) that will be used for all LLM tasks (itinerary suggestions, quotation structuring).
- The currently active provider and model (for OpenRouter, Groq, Together.AI) are displayed.
The application requires the following environment variables to be set in the .env file:
SUPABASE_URL: Your Supabase project URL.SUPABASE_KEY: Your Supabase anon key (or service role key if your RLS policies require it).GOOGLE_API_KEY: Your API key for Google AI Studio (for Gemini).GROQ_API_KEY: Your API key for Groq (for Groq models).GROQ_DEFAULT_MODEL: (Optional) Default model to use with Groq (default: 'llama3-8b-8192').OPENROUTER_API_KEY: Your API key for OpenRouter.ai.OPENROUTER_DEFAULT_MODEL: (Optional) The default model to use with OpenRouter (e.g.,google/gemini-flash-1.5,openai/gpt-3.5-turbo). Defaults togoogle/gemini-flash-1.5if not set.OPENROUTER_HTTP_REFERER: (Optional) Your site URL or app name, sent asHTTP-Refererto OpenRouter. Defaults tohttp://localhost:3000.OPENROUTER_APP_TITLE: (Optional) Your app's title, sent asX-Titleto OpenRouter. Defaults toAI Travel Quotation.TOGETHERAI_API_KEY: Your API key for Together.AI.TOGETHERAI_DEFAULT_MODEL: (Optional) Default model to use with Together.AI. Supported models includemeta-llama/Llama-3.3-70B-Instruct-Turbo-Free(default) anddeepseek-ai/DeepSeek-R1-Distill-Llama-70B-free.
The application uses the following tables in Supabase:
enquiries: Stores initial customer travel enquiries.clients: Stores client/customer information linked to enquiries.itineraries: Stores AI-generated itineraries/place suggestions for enquiries.vendor_replies: Stores vendor replies (pricing, terms) related to an enquiry.quotations: Stores final generated quotations, including the structured JSON data, links to PDF/DOCX files in Supabase Storage, and references to the itinerary/vendor reply versions used.
Refer to schema.sql for detailed table definitions and relationships.
- More sophisticated AI prompting for itineraries and quotations.
- User authentication and roles.
- Direct email integration for sending quotations.
- Advanced quotation template customization.
- Status tracking for enquiries (e.g., 'New', 'Itinerary Generated', 'Quotation Sent').
- Image integration within the PDF itinerary (e.g., destination photos).
python-docx a dependency of pdf2docx
Unexpected error fetching enquiries: Unexpected error - Type: <class 'httpx.ConnectError'>, Error: [Errno 8] nodename nor servname provided, or not known
if the above error occurs when clicking on tab2 or tab3 then the suppabase may be has to be restored.
This project is licensed under the MIT License. (Note: A LICENSE.md file is not included in the provided project files, but MIT is a common choice for such projects.)