Industrial AI Application

🛠️ The Mission

In industrial automation, data is often trapped in legacy PDF catalogs or technical datasheets. This project is a Deterministic Extraction Pipeline designed to ingest unstructured industrial PDFs and transform them into validated, machine-readable JSON schemas.

Key Technical Highlight: Unlike a standard chatbot, this system uses Pydantic-based Output Parsing to ensure that the AI follows a strict "Technical Contract." If the AI cannot find a value or tries to hallucinate a format, the system flags it during validation.

---

🚀 Getting Started

1. System Dependencies (CRITICAL)

This engine uses unstructured.io to partition complex PDF schemas. You must install the following system-level tools for the PDF partitioning to work:

• macOS (via Homebrew):

  brew install poppler tesseract

• Windows:
  1. Download and install Poppler for Windows.
  2. Download and install Tesseract OCR.
  3. Add the /bin folders of both to your System Environment PATH.

2. Python Environment Setup

# Create a virtual environment
python -m venv venv

# Activate the environment
# (Windows)
.\venv\Scripts\activate
# (Mac/Linux)
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

---

Running the Program

After all the above is complete, run the project with:

streamlit run app.py

You will be prompted to input your OpenAI product key and the associated catalog PDF.

Once those are in and processed, give it a prompt!

The output is rigid - the code is telling it only to output JSON for a motor's Model Number, Voltage, Horsepower, RPM, Size, and Enclosure.

As a proof of concept, this whole project is essentially showing that we can extract unstructured data and transform it into highly structured, readable JSON.