ERPNext Gemini Integration

Integrate Google's Gemini AI models into your ERPNext instance to enhance workflows, automate tasks, and gain insights from your data.

This app provides a framework for leveraging Gemini's capabilities, including natural language processing, multi-modal understanding (text, images, PDFs), function calling, and grounding with Google Search, directly within ERPNext.

Features

• Gemini Chat Widget: A persistent chat interface available across ERPNext (Desk and Web) for interacting with Gemini.
• Multi-modal Input: Send text prompts along with attached files (PDF, CSV, TXT, JPG, PNG) for analysis or context.
• Document Analysis: Dedicated API endpoint to analyze specific documents stored in ERPNext.
• Function Calling: Define custom functions (via Gemini Function DocType) that Gemini can call to interact with ERPNext data or perform actions (e.g., fetch data, create documents, trigger workflows).
• Conversation History: Stores and retrieves conversation history for context-aware interactions.
• Configurable Settings: Manage API keys, select Gemini models (Pro, Vision, Flash), set safety thresholds, control token limits, and toggle features via Gemini Assistant Settings.
• Audit & Feedback: Logs interactions (Gemini Audit Log, Gemini Message) and allows users to provide feedback (Gemini Feedback).
• Workflow Integration Hooks: Includes basic hooks for triggering analysis or actions on document submission or via scheduled tasks (requires further implementation).

Installation

1. Download the App:

   bench get-app https://github.com/Al-Aswany/erpnext_gemini_integration.git

2. Install the App:

   bench --site [your-site-name] install-app erpnext_gemini_integration

3. Migrate Database:

   bench --site [your-site-name] migrate

4. Restart Bench:

   bench restart

Setup & Configuration

1. Obtain a Gemini API Key: Get your API key from Google AI Studio.
2. Configure Settings in ERPNext:
   • Navigate to Awesome Bar > Gemini Assistant Settings.
   • Enter your Gemini API Key.
   • Select the desired Gemini Model (e.g., gemini-1.5-pro-latest).
   • Adjust Model Settings (Max Tokens, Temperature) if needed.
   • Configure Safety Settings (JSON format) or leave default.
   • Enable/disable Features (Grounding, Function Calling, File Analysis).
   • Set the Default Access Role required to use the assistant.
   • Save the settings.

Usage Examples

• Chat Widget: Click the robot icon (usually bottom-right) to open the chat. Ask questions, provide context, or attach files.
  • "Summarize the attached sales report PDF."
  • "What are the key points in this text file?"
  • "Based on the attached image, describe the product."
• Function Calling (Requires Defined Functions):
  • "What is the stock level for item 'XYZ-123'?"
  • "Create a draft lead for 'ABC Corp' with contact 'John Doe'."
  • "Show me my overdue tasks."

Pre-Built Functions

This integration comes with several pre-built functions to interact with common ERPNext modules:

• check_stock_levels(item_code):

  • Description: Fetches the current actual stock quantity for a specific item code across all warehouses.
  • Example Chat: "What is the stock status for item WIDGET-001?"
  • Gemini Action: Calls check_stock_levels with item_code="WIDGET-001".
  • Response: "Stock level for item "WIDGET-001" is 85."

• generate_sales_report(start_date_str, end_date_str):

  • Description: Generates a summary sales report (total orders, total amount) based on submitted Sales Orders within a date range. Defaults to the last 30 days.
  • Example Chat: "Generate a sales report for last week."
  • Gemini Action: Calls generate_sales_report (calculating dates for last week).
  • Response: "Found 15 submitted Sales Orders totaling $12,345.67 between 2025-04-22 and 2025-04-28."

• list_overdue_invoices(customer):

  • Description: Lists submitted Sales Invoices that are past their due date and not fully paid. Can optionally filter by customer.
  • Example Chat: "Are there any overdue invoices for 'Customer X'?"
  • Gemini Action: Calls list_overdue_invoices with customer="Customer X".
  • Response: "Found 3 overdue invoices for customer "Customer X" totaling $5,678.90 outstanding."

Contextual Assistance

The Gemini Assistant attempts to understand the context of the ERPNext page you are currently viewing (e.g., a specific Sales Order, Item, or Customer list).

• Automatic Context: When you open the chat widget while viewing a specific document (like Sales Order "SO-123"), the assistant is aware of this context.
• Relevant Suggestions: Based on the context, the assistant might proactively suggest relevant actions or prioritize functions related to the current document.
  • Example (Viewing Sales Order SO-123): The assistant might understand prompts like "Check production status for this order" or "Summarize this sales order."
• How it Works: The chat widget detects the current doctype and docname from the URL and passes this information along with your message to the backend.

Defining Custom FunctionsNavigate to Awesome Bar > Gemini Function > New.

2. Name: A unique identifier for the function (e.g., get_stock_level).
3. Description: A clear explanation of what the function does, used by Gemini to understand when to call it.
4. Parameters (JSON Schema): Define the input parameters the function expects in JSON Schema format.
   • Example for get_stock_level:

     {
       "type": "object",
       "properties": {
         "item_code": {
           "type": "string",
           "description": "The unique code of the item to check stock for."
         }
       },
       "required": ["item_code"]
     }

5. Enabled: Check this box to make the function available to Gemini.
6. Implementation: You will need to modify the backend code (e.g., in gemini_wrapper.py or a dedicated module) to handle the execution of this function when called by Gemini, mapping the name to your Python code that interacts with ERPNext.

Contributing

Contributions are welcome! Please follow these steps:

1. Fork the repository.
2. Create a new branch for your feature or bug fix (git checkout -b feature/your-feature-name).
3. Make your changes.
4. Commit your changes (git commit -am 'Add some feature').
5. Push to the branch (git push origin feature/your-feature-name).
6. Create a new Pull Request.

Please ensure your code adheres to existing style conventions and includes relevant tests where applicable.

License

This project is licensed under the MIT License - see the LICENSE.txt file for details.