Finding local businesses to pitch a website to was never the hard part. The hard part is deciding which of them deserve two hours of an evening. So I built the lead machine as a Claude skill: it sweeps real business data across sectors, scores each lead against an honest rubric, and drafts the cold opener for the ones that survive.
The core thesis: AI builders killed the £500 brochure site, but they did not touch businesses where the owner is time poor rather than skill poor, the ticket is high enough that a lost booking hurts, and customers research before committing. The skill hunts that shape, not a fixed list of trades.
Most of the value is in the refusing. A sector that scores brilliantly on ticket and demand often collapses on gap, because high ticket and good existing websites move together. The rubric is built to say no, and the discipline it encodes, widen sector before geography, discover broad but pitch narrow, applies to any pipeline, not just mine.