AI-Native: Live Coding an Agentic RAG That Thinks and Adapts

Live coding an agentic RAG that parses user intent into structured retrieval workflows, demonstrated through a mental‑health therapist matcher and an adaptive blogging platform.

Overview

Most AI products today are AI-enabled, adding AI to traditional databases. AI-native systems don’t just retrieve data; they think, adapt, and act.
In this live coding session, I’ll build an agentic rag that understands user intent and dynamically parses queries into structured retrieval workflows. instead of static searches, it adapts to the user’s needs.

I’ll showcase this through two ai-native products built with unbody:

An ai-native recommendation engine for mental health seekers – dynamically matching users with the right therapists.
Gray: the first ai-native blogging platform – where content structures itself on demand.

Video

Links

Tech stack