Building a Local AI Workflow with Dual A100X GPUs: RAG + Local Models + Your Company Database

Found two A100X GPUs sitting around at work and decided to put them to good use. Instead of letting them collect dust, I built a local AI system that lets employees query your company's inventory database through a conversational interface—no cloud needed, no data leaving your servers. The setup uses RAG (Retrieval-Augmented Generation) to connect a local language model directly to your inventory data, wrapped in an Open WebUI frontend so anyone can chat with it. Honestly, this was completely new territory for me, but Claude helped me figure out the heavy lifting. The whole thing actually works, and it's a solid example of how you can turn spare GPU hardware into something genuinely useful for your team.