Scoopkit is a REST API that collapses duplicate AI-industry headlines into one structured event. When 3 outlets cover the same model release or funding round, Scoopkit merges the coverage into a single event record with a real taxonomy, rather than returning 3 near-identical articles.
It's built and run by one person, not a company with a support team or a sales department. The email at the bottom of this page reaches that person directly.
The pipeline polls dozens of sources every 10 minutes, clusters near-duplicate coverage using embedding similarity and an entity-overlap check, and classifies each event into a fixed taxonomy of 10 categories and 57 subcategories. All of that is documented in detail on the blog, including 2 real bugs found and fixed in the classifier and clustering code, not just the marketing version of how it works.
Scoopkit also ships an MCP server at api.scoopkit.dev/mcp, so an AI agent can query events directly instead of a human writing REST calls by hand.
Free tier: 100 requests a day, 45-day archive, usable in production from day one. Get a key to try it.