PRISM.AI
AI DEVELOPMENT LEAD
PATTERN RECOGNITION AND INSIGHT STRUCTURE MODULE
HP AI STUDIO / MIRO / python
2025
INTRO:
PRISM is an AI-powered research synthesis tool built to solve a problem I kept running into: after user testing wraps, development stalls — not from lack of data, but from the weeks it takes to manually organize findings. I built PRISM to close that gap.
THE PROBLEM:
Research teams spend 3-4 weeks manually synthesizing data across 7 steps and 5+ platforms. Sprint after sprint gets delayed because research can't keep pace with development velocity. I experienced this firsthand and couldn't find a single tool that solved it.
RESEARCH & DEVELOPMENT
THE BUILD
STACK:
HP AI Studio · NVIDIA RTX PRO 6000 · Python · Gradio · Whisper · FigJam API · Claude.ai
BLOCKS:
A Generator-Learner feedback loop where a synthetic data engine creates realistic UX research scenarios — sticky notes, transcripts, affinity clusters — which train the pattern recognition model. The system processes multi-modal input and returns structured, source-attributed insights with confidence scoring.
PIVOT:
The original build scraped FigJam boards. Hit a wall: Figma requires auth, and scraping other researchers' data is a consent problem. Pivoted to synthetic data generation — 899,195 labeled sticky notes across 10,000 boards. Ended up being the better technical choice anyway.