Biotech Agentic Analyst

An agentic workflow for analyzing scientific papers: extract figures from PDFs using Mistral OCR 4, then run CrewAI agentic flow to generate structured intelligence — key findings, biological significance, quantitative highlights, and knowledge-base tags — for each figure.

We use:

• CrewAI for multi-agent orchestration (Flow + Crew)
• Mistral OCR 4 for PDF conversion and figure extraction
• Mistral as the LLM provider
• Streamlit for an interactive UI

How It Works

1. Upload a scientific PDF
2. Mistral OCR 4 runs with Document AI structured output to identify and extract all figures, axis labels, and captions in one pass
3. A CrewAI Flow with Analyst agent produces structured intelligence for each figure:
   • Chart type and key finding
   • Variables and conditions compared
   • Quantitative highlights
   • Biological significance
   • Knowledge-base tags
4. Results are displayed in the Streamlit UI with figure thumbnails

Set Up

Create .env File

Create a .env file in the root directory with the following content:

MISTRAL_API_KEY=<your_mistral_api_key>

Install Dependencies

uv sync
source .venv/bin/activate

On Windows (PowerShell):

uv sync
.venv\Scripts\activate.ps1

Run the Streamlit App

streamlit run app.py

Open the URL shown in the terminal (e.g. http://localhost:8501). Upload a scientific PDF in the sidebar and click Run Analysis to trigger the full workflow.

📬 Stay Updated with Our Newsletter!

Get a FREE Data Science eBook 📖 with 150+ essential lessons in Data Science when you subscribe to our newsletter! Stay in the loop with the latest tutorials, insights, and exclusive resources. Subscribe now!

Contribution