Prophecy + Claude Skills: Automating Data Workflow Migration with AI (Demo)
Rebuilding a legacy report by hand can take days. In this demo, Callum Finlyson shows how AI Skills change that: reusable instruction packages that teach Claude exactly how to handle a specific, multi-step data task, the same way every time. Watch Prophecy take an existing Axiom report (a proprietary financial reporting tool) straight from its XML export and rebuild the entire logic inside Prophecy. Using a custom "Axiom Flow Explainer" skill, the agent reads the exported files, pulls context from Prophecy's data catalog, and recreates a daily deposits pipeline as governed, DBT Core-compliant models running on Databricks SQL. What you'll see: - What AI Skills are: markdown-based instructions that teach Claude to complete a complex task consistently, without you re-prompting every time - How Prophecy imports proprietary formats like Axiom XML and turns them into deterministic, production-ready pipelines - Why visual inspect-and-refine matters: query the output of every transformation step to confirm the AI got it right, before you trust the result - How the finished pipeline compiles to open DBT Core and SQL and runs on Databricks, no proprietary format required Prophecy is the AI data prep platform for business analysts, data engineers, and platform teams who need governed, production-ready pipelines without the engineering bottleneck. Prophecy pairs Claude with a visual workflow you can inspect, correct, and trust. 📌 Try Prophecy: https://www.prophecy.ai/ 📌 Connect with Callum: https://www.linkedin.com/in/finlaysoncallum/ Chapters: 0:00 – Intro: Callum on Prophecy AI skills 0:11 – What AI data prep looks like in Prophecy 1:06 – What AI skills actually are 2:02 – The demo: recreating an Axiom report 2:58 – What is Axiom? 3:23 – Creating the project and installing the skill 3:48 – Running the Axiom Flow Explainer skill 4:11 – Axiom logic converted into Prophecy 4:21 – Behind the scenes: DBT Core models 4:34 – Running the rebuilt pipeline 5:01 – Live on Databricks SQL with DBT Core 5:15 – Inspecting every transformation step 5:47 – A deterministic, governed pipeline