ThruWire uses Jev classifiers for per-sentence meeting checkpoints

Josh Rosen @JoshARosenOriginally posted 2026.09.19

Open original on X

Compounding use case for Jev. What we did: - defined checkpoints in ThruWire at every sentence/moment boundary within each granola conversation - structured it as a graph of meeting sections - had the agent harness push all the artifacts to the ThruWire checkpoint system to kick off each Jev classifier - used ThruWire custom renderer feature to display the data nicely.

This post quotes another post: “Jev added to Granola and ThruWire for meeting sentiment analysis”