THE DHRUV XORGGPT04 / 13

Technical Feasibility & Solution Approach

← All project documents

I have been taking guidance from Professor Emaad Manzoor (who teaches AI for Business Applications at Cornell) to refine the technical approach.


Key Technical Strategy: Retrieval-Augmented Generation (RAG) Training

The recommended solution leverages RAG training, where the effectiveness hinges on data collection, our secret sauce. Unlike traditional models that rely solely on pre-existing documentation, this approach integrates real-time university communications to maintain information accuracy. Along with the current documentations and FAQs, the system will automatically monitor staff emails, with an option for privacy (allowing staff to disable monitoring for sensitive emails). This filtered data would then be fed into the model. This real-time ingestion process minimizes outdated responses and improves the model’s context-awareness.


The Secret Sauce

This isn’t just an AI product, because the AI part (RAG training) is relatively straightforward to implement. The real challenge and innovation lie in data collection, refinement, and optimization. After conducting numerous customer interviews across different universities, I have accounted for multiple data challenges and caveats, ensuring that the system effectively adapts to each university’s unique workflows. The true differentiator is how the system structures, filters, and applies the right data, making orgGPT fundamentally more effective than any other solutions.

Continue readingUser Personas

Open original ↗
THE DHRUV XPERSONAL PORTFOLIO
BEFORE THE FIRST PAGE

Strike a
match.

Swipe left or right across the strip to light the edition.

Move your light. Explore the paper.Toggle it anytime in the header.