📊 Extension Impact Narrative Agent

Transform your raw Extension data into professional impact narratives and visual dashboards. This application provides a multi-agent AI pipeline designed to automate the heavy lifting of county-level reporting for Extension professionals.

✨ Features

• Dynamic Impact Dashboard: Readily visualize KPIs like Total Contacts, Knowledge Gain, and Delivery Methods via interactive Plotly charts.
• Intelligent Filtering: Slice your data seamlessly by subsets (e.g., County or Topic) to calculate hyper-specific insights.
• Multi-Model LangChain AI Pipeline:
  • Agent 1 (Narrative Engine): Automatically processes filtered statistical metrics and qualitative feedback into a comprehensive, written UF/IFAS-standard impact report.
  • Agent 2 (Interactive Q&A): A context-aware chatbot grounded exclusively in your dataset, ready to answer questions, write emails, or pull specific insights on demand.
• 1-Click Comprehensive PDF Export: Generates a perfectly formatted PDF spanning your dashboard visualizations, the structured core narrative, and the full interactive Q&A history.

---

🚀 Getting Started

Follow these steps to set up and run the application locally.

1. Clone the Repository

Download the project to your local machine:

git clone https://github.com/siddy819/Extension-AI-Agent.git
cd Extension-AI-Agent

2. Set Up a Virtual Environment (Highly Recommended)

Creating a virtual environment ensures that the required packages don't conflict with other projects on your system.

On Windows:

python -m venv venv
venv\Scripts\activate

On macOS/Linux:

python3 -m venv venv
source venv/bin/activate

3. Install Dependencies

Install all the required Python libraries using the generated requirements.txt file:

pip install -r requirements.txt

4. Run the Application

Boot up the Streamlit server:

streamlit run app.py

The app should automatically open in your default web browser at http://localhost:8501/.

---

🗂️ How to Use the App

1. Provide API Credentials: In the left sidebar, enter the secure API keys for Agent 1 and Agent 2 (powered via the UF IT enterprise endpoint).
2. Upload Data: Drag and drop the provided mock_data.csv file (or your own CSV data matching the exact schema) into the upload box.
3. Filter & Explore: Use the dropdown menus to refine which Counties or program Topics you want to analyze.
4. Generate the Report: Click "🚀 Run Agent 1" to watch the narrative stream onto the screen.
5. Ask Follow-Ups: Scroll down to Agent 2's chat bar to ask specific, grounded follow-up questions about the data subset.
6. Export: Click the "💾 Download Comprehensive Report (.pdf)" button to save the entire end-to-end workflow to your hard drive.

---

📝 Required Data Schema

If you use your own dataset, the uploaded CSV must contain exactly these headers: Program Date, County, Program Name, Topic, Total Contacts, Indirect Reach, Surveys Collected, Knowledge Gain (%), Behavior Change (%), Program Type, Target Audience, Success Story, Qualitative Feedback