Dino Detective
Upload a photo - a fossil, a toy, a footprint, even something delightfully unrelated - and let an AI agent identify it. Claude's vision model profiles the specimen (species, period, diet, size, confidence, fun fact), then hands over to a contextual chat: every follow-up question is answered with the identified specimen in mind, whether it's a Brachiosaurus or a SoftBank robot.
Under the hood: a fully agentic flow on Lucee 7's native AI integration - multimodal image analysis (Claude), fast free-tier conversation (Groq), structured JSON output, HTMX out-of-band swaps for a smooth no-reload UI, and AI providers switchable from a single line in `Application.cfc`. No Python, no Node backend, just CFML doing real agentic work.
A small proof that legacy platforms can still play in the AI era.
Test project
Hi, I’m JC Royer!