Automate Data Prep with Specialized Claude Code | Databricks Data + AI Summit 2026

What if your analysts could build production-ready data pipelines in minutes without writing a single line of code? In this session from Databricks Data + AI Summit, Nathan Tong, a lead forward-deployed engineer at Prophecy, shows how Prophecy's specialized AI agent (built on Claude) lets business users prep data on Databricks as fast as a data engineer. You'll also see how enterprises can migrate legacy ETL workflows from tools like Alteryx, Informatica, or DataStage in a single import. The core problem: business users need data, data engineers build data pipelines, and the back-and-forth cycle takes weeks. Desktop tools like Alteryx get analysts unblocked, but they create data copies outside the governed cloud environment. Prophecy bridges the gap with a no-code, AI-native data prep interface that runs natively on Databricks, generates open-source dbt SQL, and is version-controlled on Git. You'll watch two live demos: Demo 1: Build a pipeline from a prompt Demo 2: Import a legacy ETL workflow Key topics covered: • Why the business user/data engineer handoff creates weeks of delay • Why generic Claude isn't enough, and what a specialized agent adds • How Prophecy generates standardized, deterministic, testable SQL from natural language • Unity Catalog integration: agents that only access governed data sources • Open-source dbt core: no vendor lock-in, runs on Databricks, BigQuery, Snowflake • Visual pipelines with step-by-step explainability and built-in unit testing • Legacy ETL migration: importing Alteryx, Informatica, Ab Initio, and DataStage workflows • Bidirectional code ↔ visual editing for analysts and engineers working side by side 🔗 Learn more about Prophecy: https://www.prophecy.ai Timestamps 0:00 – Introduction 1:13 – The Data Prep Problem 2:26 – Introducing Prophecy 3:59 – Demo 1: Building a Pipeline with AI 12:41 – Demo 2: Migrating from Legacy ETL 16:40 – Summary & Key Takeaways