Solving and Alteryx Challenge in Minutes: Prophecy Agentic Data Prep Demo

How long does it take to go from a business question to a working data pipeline? With Prophecy, just a couple of prompts and a few minutes. In this demo, watch Nathan Tong tackle a real Alteryx challenge: figuring out which bank failed fastest using two sample datasets. What you'll see: - A single prompt turns into a full pipeline, with no manual mapping of data sources. The Prophecy agent scans Unity Catalog (and external databases, file shares, even SharePoint) to find what it needs on its own - Why a specialized agent trained on Claude beats prompting Claude directly for data work: consistent, repeatable results instead of 10 different answers to the same question - Every step in the pipeline shown visually as gems (Alteryx users will know them as tools), so you can verify exactly how the answer was built, not just trust a black box - A second prompt adds a bar chart and pie chart breakdown by acquiring institution and region, with the agent automatically generating unit tests to confirm each step produces a real result - A look under the hood at the open-source DBT Core and SQL code Prophecy generates, no DBT Cloud fees, no vendor lock-in, and portable across Databricks, BigQuery, Redshift, and more The result: two prompts, about a minute and a half, and a fully governed, inspectable, production-ready pipeline with a live dashboard attached. Prophecy is the agentic data prep platform for business analysts, financial auditors, portfolio managers, and data teams who need governed, production-grade pipelines without the engineering bottleneck. 📌 Learn more at https://www.prophecy.ai/replace-alteryx Timestamps 0:00 — The challenge: bank failure analysis 0:59 — One prompt into the Prophecy agent 1:09 — Why a specialized agent beats prompting Claude directly 1:46 — How the agent finds data sources on its own 3:03 — Gems: the visual, verifiable steps behind every answer 4:08 — Prophecy explains the pipeline step by step 4:31 — The challenge answer 4:42 — Inspecting the data and checking for quality issues 5:10 — Second prompt: bar chart, pie chart, and regional grouping 5:55 — Auto-generated unit tests confirm every step works 6:31 — Under the hood: open-source DBT Core and SQL 7:23 — The finished dashboard, live against the warehouse 7:42 — Recap: governed, inspectable, production-ready in minutes