Multi-Modal AI Chat Platform with Excel Export & Document Processing
A powerful AI assistant powered by Google's Gemini 2.0 Flash that combines conversational AI with advanced document processing, image analysis, and automatic Excel export capabilities for data-driven workflows.
Check out the live app here 👉 GeminiFlow on Streamlit
- Multi-Turn Conversations: Context-aware responses with conversation history
- Streaming Responses: Real-time text generation with typing animation
- Quick Prompts: Pre-configured prompts for common tasks
- Session Management: Track conversation duration and message count
- Smart Context Handling: Maintains up to 5 previous messages for context
- Automatic Table Detection: Identifies markdown tables in responses
- One-Click Excel Download: Instant conversion to formatted .xlsx files
- Smart Formatting: Auto-adjusts column widths based on content
- Professional Styling: Clean, organized spreadsheets ready for business use
- Raw Data Access: Copy markdown tables directly
- Multi-Format Support: PNG, JPG, JPEG, WebP, GIF
- Smart Resizing: Automatic compression for large images (max 4096px)
- Visual Preview: In-sidebar image display with dimensions
- Data Extraction: Extract text, numbers, and patterns from images
- Chart Analysis: Understand graphs, diagrams, and visual data
- Multi-Page Support: Handle documents of any length
- Progress Tracking: Real-time extraction progress for large files
- Smart Truncation: Optimizes long documents for AI processing (8000 chars)
- Word Count Stats: See pages and word count instantly
- Text Extraction: Pull content from complex PDFs
- Temperature Control: Adjust creativity (0.0-1.0)
- Token Limits: Control response length (256-8192 tokens)
- Model Settings: Fine-tune AI behavior
- Export Options: JSON, Markdown, Excel formats
- Chat History (Markdown): Complete conversation in .md format
- JSON Export: Structured data with timestamps
- Excel Tables: Individual table downloads
- Professional Formatting: Ready-to-share reports
- Python 3.8 or higher
- Google API Key (Gemini 2.0)
- pip package manager
- Clone the repository
git clone https://github.com/Brijesh1656/GeminiFlow.git
cd GeminiFlow- Install required packages
pip install streamlit google-generativeai python-dotenv Pillow PyPDF2 pandas openpyxl- Set up your API key
Create a .env file in the project root:
GOOGLE_API_KEY=your_google_gemini_api_key_hereOr use Streamlit secrets (.streamlit/secrets.toml):
GOOGLE_API_KEY = "your_google_gemini_api_key_here"Get your API key: Google AI Studio
streamlit run app.pyThe application will open in your default browser at http://localhost:8501
- Type your message in the chat input at the bottom
- Press Enter or click Send
- Watch AI respond in real-time with streaming text
- Images: Use sidebar to upload PNG, JPG, GIF, WebP
- PDFs: Upload documents for summarization or analysis
- Files attach automatically to your next message
- Click Quick Prompts in sidebar for templates
- Customize the prompt with your specific data
- Great for Excel tables, math problems, analysis
- Ask for data in "markdown table format"
- AI generates formatted tables automatically
- Click "📥 Excel" button below response
- Download instantly as .xlsx file
- Temperature: Higher = more creative, Lower = more focused
- Max Tokens: Control response length
- Experiment to find your perfect settings
- Chat History: Download full conversation (Markdown/JSON)
- Individual Tables: Excel files for each data table
- Raw Data: Copy markdown for other tools
| Technology | Purpose |
|---|---|
| Streamlit | Web application framework and UI |
| Google Gemini 2.0 | Advanced multi-modal AI model |
| PyPDF2 | PDF text extraction and processing |
| Pillow (PIL) | Image processing and optimization |
| Pandas | Data manipulation and DataFrame creation |
| OpenPyXL | Excel file generation and formatting |
| Python-dotenv | Environment variable management |
- Convert text data to Excel-ready tables
- Extract numerical data from images/PDFs
- Generate formatted financial reports
- Step-by-step math solutions with tables
- Physics/chemistry calculations
- Engineering computations with formulas
- Summarize long PDFs quickly
- Extract key points from reports
- Analyze contracts and legal documents
- Extract data from charts and graphs
- Analyze infographics and diagrams
- OCR-like text extraction from images
- Generate professional data tables
- Create formatted Excel reports
- Analyze financial statements
- Process invoices and receipts
This application is for educational and productivity purposes.
- Requires valid Google Gemini API key
- Subject to Google's API usage policies and quotas
- AI responses should be verified for critical applications
- Not a substitute for professional advice in specialized fields
- Large PDFs: Files over 100 pages may take time to process
- API Quotas: Free tier has rate limits (handle gracefully)
- Complex Tables: Some nested tables may need manual adjustment
- Image Quality: Low-resolution images may reduce accuracy
- Voice input/output support
- Multi-language conversation support
- Code execution environment
- Advanced chart visualization
- Collaborative chat sessions
- Custom prompt templates library
- Integration with Google Drive
- Mobile-responsive design
- Chat branching and versioning
- API endpoint for external apps
Brijesh Singh
- GitHub: @Brijesh1656
- LinkedIn: brijesh-singh-b84275307
- Email: brijesh7146@gmail.com
- Location: Hyderabad, India
BBA (Business Analytics) student passionate about leveraging AI and data science to solve real-world problems. Experienced in Python, machine learning, and building intelligent applications.
Check out my other AI and data science projects:
- FinVision - AI-powered financial document analysis with risk detection
- Stock Analysis Pro - Technical analysis platform with AI insights and backtesting
Contributions are welcome! Here's how you can help:
- Fork the repository
- Create a feature branch (
git checkout -b feature/AmazingFeature) - Commit your changes (
git commit -m 'Add some AmazingFeature') - Push to the branch (
git push origin feature/AmazingFeature) - Open a Pull Request
- Follow PEP 8 style guide
- Add comments for complex logic
- Test with various file types
- Update documentation for new features
This project is licensed under the MIT License - see the LICENSE file for details.
- Google for Gemini 2.0 Flash API
- Streamlit community for excellent framework
- Open-source contributors
- All users providing feedback and suggestions
For support, questions, or feedback:
- Email: brijesh7146@gmail.com
- Issues: Open an issue on GitHub
- LinkedIn: Message me directly
- Be Specific: "Create a table with columns: Name, Age, Score" works better than "make a table"
- Request Formats: Explicitly ask for "Excel-ready format" or "markdown table"
- Step-by-Step: Add "show calculations step-by-step" for math problems
- Context Matters: Upload PDFs/images before asking questions about them
- Tables must use markdown format with
|separators - Include a header row with
|---|---|separator - AI automatically formats for Excel compatibility
- Download appears immediately below table responses
- Images: Under 10MB work best
- PDFs: First 8000 characters used (usually 10-15 pages)
- Multiple Files: Upload one of each type (1 image + 1 PDF)
⭐ Star this repo if you find it helpful!
Made with ❤️ and ☕ by Brijesh Singh