Turning messy web signals into a repeatable operating system
Grubhub’s merchant analytics team already had a strong scoring model, but it systematically missed restaurants that mattered culturally and locally. Those signals lived in fragmented, messy sources like Reddit, YouTube, and local review communities.
TinyFish made those sources usable in production. Agents continuously traversed discussions, resolved restaurant entities across noisy mentions, and returned structured outputs the team could incorporate into its merchant scoring system.
We have a lot of data already, but we were missing genuine crowd-sourced sentiment. TinyFish fills that intelligence gap really well.
