Usage

Starting the Platform

# Activate virtual environment
source .venv/bin/activate

# Start the API
uvicorn app.api.main:app --reload

# In a second terminal, start the UI
python ui/gradio_app.py

Open your browser at http://localhost:7860

Querying Documents (RAG)

Type any question related to your research documents :

"What are the latest findings on skin microbiome and cosmetic formulation?"
"Summarize the safety assessment for UV filter X"
"What does the OECD report say about nuclear waste management?"

The RAG engine will retrieve the most relevant chunks from SharePoint and generate a sourced answer.

Querying Analytics Data

Ask data-driven questions in natural language :

"What is the average experiment success rate per research team?"
"Show me the top 5 compounds tested in Q1 2025"
"Compare results between lab A and lab B"

The Analytics Agent translates your question into pandas operations and returns structured insights.

Combined Queries

ResearchMind shines when combining both sources :

"Based on the latest reports and our experimental data,
which compounds show the most promise for UV protection?"

The Supervisor Agent automatically routes to both tools and synthesizes a unified answer.

Ingesting New Documents

New documents added to SharePoint are automatically ingested on the next pipeline run :

python -m app.rag.pipeline