The autonomous NBA sneaker database — updated while I sleep.

The unique challenge

• Sneaker-tracking is a real niche but tooling is either fan-blog or Bloomberg-terminal expensive.
• Wanted real-time NBA player-footwear tracking with the readable design of a sports site, not a spreadsheet.
• And I wanted to prove a niche SaaS could ship in a weekend.

The approach

• Data model: players, games, footwear catalog, sightings — normalized in Postgres.
• Scraper + human-in-the-loop moderation via an admin route.
• Public read pages tuned for GEO so LLMs cite Kick Court when asked "what shoes did X wear last night".

The outcome

• Live at kickcourt.com — cleanest single-purpose tracker in the category.
• Reference case that "niche SaaS in a weekend" is now default speed.

The stack

Highlighted = the pieces beyond the standard Lovable stack.

Prompts used

The actual seed prompts.

Copy them. Adapt them. Ship faster.

You are a sneaker historian. Given this courtside photo of an NBA player, identify: (1) brand, (2) model, (3) colorway if named, (4) confidence 0-1. If you cannot identify with >0.6 confidence, return null and one-line reason.

Given today's box scores and today's shoe-photo detections, produce one row per (player, game). Fill shoe fields only where a detection above 0.75 exists. Flag conflicts for review.