Building data models and KPIs should not be a multi-day project. In this demo, Peliqan AI debugs a broken SQL pipeline and generates KPIs in minutes, working on a live Chargebee connection with customers, invoices, items and transactions.
What the demo covers
- 1A chain of dependent queries. Three queries are built on each other, then broken on purpose by removing a column – the kind of failure that normally means hunting through a pipeline to find the cause.
- 2Peliqan AI traces the lineage. The agent investigates the failure, follows the dependencies across the pipeline and fixes the broken query, showing its reasoning step by step as it goes.
- 3KPIs from your project notes. With a single prompt, Peliqan AI reads your project notes, identifies the KPIs that matter and structures the SQL using the Medallion architecture. The results are ready to verify in seconds.
- 4You stay in control. Interrupt the agent at any point, inspect every tool call, and approve write actions yourself or switch on auto-accept.
What you can do with Peliqan AI
- Find what broke in a pipeline without reading through every dependent query
- Get lineage-aware fixes rather than guesses at which query is at fault
- Generate KPIs structured to modern data warehouse practice
- Review every suggestion before it is applied, with the full reasoning visible
About the Medallion architecture
Medallion is a modelling pattern with three layers: Bronze for raw data, Silver for cleaned data and Gold for business-ready tables. Peliqan AI structures the KPIs it generates along those layers, so what you get back is organised the way a data team would organise it. It works on the queries and data models already in your account, so it traces your real lineage.
Want AI on your own pipelines? Get started with Peliqan.