This is such a great initiative, because "DATA" is the moat now.

@Tiny_Fish launched a 15-company alliance to bring specialist partner datasets closer to AI agents.

Its founding partners include Crunchbase, Similarweb, Databento, RocketReach and Trellis, covering 10 categories from finance to travel.

A model can search the public web all day, but it cannot reconstruct private-company intelligence, verified identity, market feeds, or court data that lives behind licensed sources.

Model weights are copied, distilled, replaced and rerouted constantly. Proprietary datasets are much harder to reproduce.