Latest blog posts by Paweł Szczepanik

AI governance tools buyer's guide: seven criteria, from model inventory and policy controls to audit evidence, stack fit, EU AI Act mapping and licensing, checked against vendor documentation.

November 18, 3:00 PM CET, online and in Polish: how Databricks, DS Stream and PayPo build business context for AI, with a live Metric Views demo.

How to build a Databricks semantic layer with metric views: a worked YAML example, where each rule belongs, and how to test Genie against gold SQL.

Markdown optimization retail guide: which SKUs to mark down, when, and how deep, how a sell-through target drives the model, and what Zara, Walmart and ASOS reported from field tests.

Watch a short DS Stream demo of AI product intelligence that prices new products with no sales history by matching them to similar SKUs.

Model Context Protocol enterprise guide: what MCP standardizes between agents and your systems, how it differs from function calling and RAG, and how to deploy it with OAuth, a gateway and a registry.

Data warehouse design best practices for the AI era: declare the grain, keep a dimensional core, version history for reproducible training sets, and define metrics where AI agents can read them.

Explore a Minesweeper AI case study: deterministic rules, agent memory, guardrails, and a 2,520-game test of reliable decision-making.

Watch a five-minute DS Stream demo of AI product segmentation that sorts thousands of SKUs into margin drivers, traffic drivers, and inventory risk.

An AI POC checklist built as five gates, each with an artifact, an owner and a pass condition, ending in a go, stop or re-scope decision.