Introduction
This week, our team went deep on both technical implementation and user experience design. We're moving from concept to concrete prototypes, understanding how vector embeddings work while simultaneously mapping out user flows and wireframes.
Design Team Updates
Reagan continued working on lo-fi designs and user flows, receiving valuable feedback from Henry about the hi-fi direction. Bruno started documenting user personas, creating detailed profiles of our target users to guide design decisions. Darrel completed a competitive analysis and finished survey edits, then began working on user flow diagrams to map the journey through our application.
The design team is building a comprehensive understanding of who will use Panday and how they'll interact with it.
Development Team Updates
Nikita made significant progress understanding vector embeddings, reading an excellent Cloudflare article on the topic and exploring pgvector Docker images, embedding model leaderboards, and Neon's AI concepts documentation. Ozem created a mind map using ReactFlow framework to visualize the project structure, showing how clicking timeline nodes could display generated roadmaps or job requirements.
Josh analyzed the SkilledTradesBC website with Claude AI, discovering that while the content is comprehensive, the navigation is challenging—validating our mission to make this information more digestible. Peter crawled the SkilledTradesBC electrician construction site, extracting approximately 3,200 JSON files and uploading them to GitHub for processing.
TL;DR
We deepened our understanding of vector embeddings and began implementing them with pgvector. The design team created user personas and flows while developers built prototypes and gathered data from skilled trades resources.
What's Next?
Week 4 will see us deploying our first embedding API, refining our visual designs based on feedback, and beginning to connect our frontend prototypes with backend services.