Week 8: RAG Integration and User Profiles

Introduction

This week marked a significant milestone: integrating our RAG (Retrieval-Augmented Generation) agent into the live application. Panday can now intelligently answer questions using our curated knowledge base while staying on topic and admitting when it doesn't know something.

Design Team Updates

Reagan checked over the current tutorial and roadmap, brainstorming ideas to increase UX flow. She started working on guest and profile views, sign-in flows, and began designing the home page and profile page. Bruno planned the presentation structure: team intro, problem and passions, solution demonstration with tutorial, future prospects, and additional app features.

Darrel designed the landing page for Panday and worked on brochure design, completing edits based on team feedback.

Development Team Updates

Nikita successfully implemented the LlamaIndex RAG agent into the app with a test document, demonstrating good performance at staying on topic and acknowledging knowledge limitations. He updated the UI and switched from the FastAPI server with SentenceTransformers to OpenAI's text-embedding-3-small model, creating a working prototype.

Ozem got the tutorial system fully functional and began improving resource and roadblock checklist nodes to display helpful information. Josh helped Darrel with hi-fi designs and pondered the future of voice-assisted AI for Panday—imagining conversational interfaces that spawn nodes and information dynamically.

Peter added chat history functionality and FAQ integration to the chat interface, allowing users to click FAQ questions to populate the chat input. He created a chat-history-merge branch with basic functionality published.

What's Next?

Week 9 will focus on collecting and indexing more documents, improving node information displays, and finalizing our marketing materials for submission.