The AI Car Wash: Transforming Dormant Media into Intelligent Assets

At IBC I presented the “AI Car Wash,” a tool we built to solve a problem that every media company faces: thousands of hours of video content sitting in archives, unindexed, untagged, and effectively invisible. This content represents enormous latent value, but without proper metadata it cannot be searched, licensed, or monetized. The AI Car Wash takes raw video files and runs them through an automated pipeline that extracts rich metadata, generates IAB advertising tags, creates scene-level descriptions, and builds a semantic search index.

The most compelling part of the story is how we built it. A small team of generalists, armed with AI development tools, went from concept to working prototype in just 14 days. This is the kind of speed that was simply not possible before the current generation of AI tooling. We did not need a team of machine learning engineers or a six-month roadmap – we needed curiosity, domain knowledge, and the right AI-powered workflow.

The end result is a system where editorial teams can query their entire archive using natural language. Instead of scrolling through spreadsheets of timecodes, they ask questions like “find all segments featuring underwater photography from 2019” and get accurate, timestamped results. We also integrated MCP connectors so that AI assistants can query the enriched archive directly, opening up entirely new workflows for content discovery and repurposing.