Project Catalog
AI-Powered Political File Management System
Problem
During the 2024 election cycle, designers and account managers were spending 1–5 minutes per file search in Dropbox. With hundreds of daily file requests, this was costing the team hours each week. The deeper issue was that Dropbox only searches by file name or exact keywords — there was no way to discover past creative work by theme, message, or visual concept.
Approach
I designed and built a full-stack AI-powered catalog that automatically analyzes and semantically tags political design files. The system uses vector embeddings and multi-modal AI to understand the content of each file, enabling natural language queries like "healthcare policy graphics" or "Biden campaign materials." It integrates with existing Dropbox workflows so the team didn't need to change how they store files.
Outcome
Reduced file discovery time from 1–5 minutes per search to near-instant results. Designers gained the ability to find creative inspiration from past campaigns by theme, candidate, or concept — not just by file name. Account managers could fulfill requests in seconds rather than manually navigating folder hierarchies.