Challenge 3

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

What I Built

Industrial AI Assistant — Your Factory's Knowledge at Your Fingertips

Industrial teams often waste valuable time searching through lengthy SOPs, equipment manuals, and technical documents to find answers to simple but critical questions.

I am building a Multimodal GenAI Assistant that makes industrial knowledge accessible through a simple conversational interface.

Users can ask questions in natural language and get clear answers grounded in approved industrial documentation, along with source references to help them verify the information. The long-term goal is to support scanned documents, diagrams, and equipment images as well.

Who would pay for it?

• Manufacturing companies and factories
• Maintenance and engineering teams
• Industrial plants and equipment operators
• Organizations managing large technical documentation libraries

The business opportunity is straightforward: reduce the time spent searching for information, make documentation easier to use, and help teams access relevant knowledge when they need it.

Demo

Demo coming soon.

Code

GitHub repository: Coming soon.

How I Built It

The project is designed around Retrieval-Augmented Generation (RAG), combining document retrieval with generative AI.

The planned workflow is:

1. Upload industrial SOPs, equipment manuals, and technical documents.
2. Extract text from PDFs and scanned pages using document parsing and OCR.
3. Divide the content into searchable sections and create embeddings.
4. Retrieve relevant sections when a user asks a question.
5. Use a language model to generate an understandable answer grounded in the retrieved information.
6. Display source references so users can verify the answer against the original documentation.

Planned technology stack:

• Frontend: React
• Backend: Python and FastAPI
• AI architecture: RAG and embeddings
• Knowledge retrieval: Vector search with PostgreSQL/pgvector
• Document processing: PDF parsing and OCR
• AI models: An open-weight language model evaluated for accuracy, hardware requirements, and deployment cost

The focus is not just on generating answers. It is on making those answers traceable to the documentation they come from and acknowledging when the available evidence is insufficient.

Why Does Open Innovation Matter?

Industrial AI should not be limited to organizations that can afford expensive proprietary systems.

Open innovation makes it possible to experiment with open-weight models, customize retrieval pipelines, inspect system behavior, and deploy components in environments with different privacy, cost, and infrastructure requirements.

For this project, open-source tools can help reduce vendor lock-in, make the architecture more adaptable, and provide a path toward private deployment for organizations that cannot send confidential manuals to external services.

It also creates an opportunity for developers to contribute improvements in document parsing, multilingual retrieval, evaluation, and safety mechanisms.

My goal is to build something practical that can evolve through community feedback rather than remain a closed demonstration.

My Agent Session

Not available yet.

Prize Categories