🧠 ADE Helper Scripts

This repository contains helper scripts, pipelines, and demo applications built around LandingAI’s Agentic Document Extraction (ADE). The goal is to showcase document understanding workflows across different domains (financial filings, CME certificates, batch processing apps) and make them reproducible for internal and partner teams.

⚠️ Disclaimer

All sample helper scripts are provided “as is” for use with Agentic Document Extraction (ADE).
No warranty is expressed or implied, and no support is provided.

---

📁 Repository Structure

ade-helper-script/
├── Industry_Use_Cases/
│   └── FinServ_Scripts/
│       ├── ADE_10K_Pipeline_Local/    # Local ADE + RAG pipeline for SEC 10-K filings
│       └── Edgar_API_Pipeline/        # Extracts and processes 10-Ks from SEC EDGAR
│
├── Workflows/
│   └── Streamlit_Application_Batch_Processing/
│       └── app.py                     # Webinar demo Streamlit app for batch document parsing
│
├── requirements.txt
├── .env.example
└── README.md

---

🚀 Projects in This Repo

1. Industry Use Cases

• Financial Services Pipelines

  • ADE 10K Pipeline Local

    • Extracts financial metrics from 10-K filings.
    • Runs ADE locally, saves outputs (markdown, JSON, visual grounding).
    • Generates embeddings with OpenAI, stores in ChromaDB, supports RAG (Retrieval-Augmented Generation) querying with visual grounding.

  • EDGAR API Pipeline

    • Downloads 10-Ks directly from the SEC EDGAR API.
    • Prepares filings for ADE processing and downstream analysis.

🌐 Deployment on AWS (Optional)

For production, you can extend these workflows into a serverless, event-driven pipeline:

• S3 – upload documents to trigger processing.
• Lambda – run ADE parsing logic.
• OpenSearch – store and query embeddings at scale.
• Bedrock (Claude, Titan) – perform LLM-based RAG Q&A.

---

2. Workflows

• Streamlit Batch Processing App
  Interactive web app (from LandingAI webinar) that:

  • Selects a local folder of documents.
  • Runs ADE on all PDFs/images.
  • Displays JSON outputs and page-level bounding-box visualizations.
  • Supports resetting/re-running without restarting the app.

  Run it with:

  streamlit run Workflows/Streamlit_Application_Batch_Processing/app.py

---

📚 Learn More

• LandingAI ADE Documentation
• Agentic Document Extraction Playground