I Built a Live AI App in 30 Minutes: Postgres → Estuary → Webhook → Jev

What if every new row in your database could be classified by AI the moment it's created, with no batch jobs and no waiting? This video is sponsored by Estuary: https://estuary.dev/ In this video, I show a live app that does exactly that. Data streams from a local Postgres database through Estuary, which captures the changes and materializes them via webhook into an application. As each cart update arrives, Jev, a System One model from typesafe.ai, scores the data and returns probabilities for the labels we define. The result is a real-time view of what a grocery shopper is building, like a meal forming as items are added to the cart. We'll cover: - What Jev is and how it differs from text-generating LLMs - Capturing from Postgres with Estuary and sending data to a webhook materialization - How the app calls the Jev API and updates the page as data arrives - Why this pattern matters for personalization, follow-up, and discounts Resources: - Project repo: https://github.com/early-signal-tech/estuary-postgres-webhook-jev-stream - Estuary: https://estuary.dev/ - Jev: https://typesafe.ai/ - Estuary unbundled DW video: https://www.youtube.com/watch?v=mtq63fDyqtA - Transform Streaming Postgres: https://www.youtube.com/watch?v=hzA55z52YdE Prerequisite: Docker installed. If you enjoyed this, please like and subscribe. It helps the channel grow. My Links: Substack: https://thefulldatastack.substack.com/ LinkedIn: https://www.linkedin.com/in/hoytemerson/ X: https://x.com/HoytEmerson My content company: https://www.earlysignal.tech/