Intro to Steep

In this live demo and Q&A, Sebastian Klintberg from our commercial team walks through what Steep is and how it looks and feels for end users, and Oscar Albrecht, one of our senior software engineers, takes us under the hood to show how the semantic layer that powers it all gets built. Watch the full recording below.

What we covered

• Why Steep flips the traditional BI model: data and analytics teams govern the semantic layer, metrics, dimensions, and entities, while everyone else gets to explore, build reports, and run AI analysis on their own, without waiting on ad hoc requests.
• A tour of the platform: the home screen, the metrics catalog, exploring and breaking down metrics, building reports (Steep's take on dashboards), drilling into row-level entities, and using Steep AI to analyze data and draft reports from a prompt.
• How the semantic layer gets built: defining modules, dimensions, metrics, and entities directly in the UI, or as code synced from a GitHub repo, and how the two approaches work together in the same workspace.
• Real customer results: teams like Voi went from 10-20% monthly active usage with legacy BI tools to as high as 90% with Steep, while cutting down the number of one-off dashboards their data teams had to maintain.

From the Q&A

• How do you stop different teams from ending up with different definitions of the same metric? Every metric is defined once in the semantic model. Whether you're exploring it in Steep, asking Steep AI, or querying through the MCP server from Claude or ChatGPT, you always get the same definition.
• Can the AI generate its own SQL that might contradict the semantic layer? No. Steep AI doesn't have direct access to your data warehouse and can't fire arbitrary SQL against it. It can only query through the governed semantic layer, so answers are always grounded in the same definitions.
• Do teams replace dashboards entirely, or use both? Most teams keep both. Reports still have a place for the numbers you check on a recurring basis, but unlike traditional dashboards, they have no logic baked in. They're built from the same governed metrics, so a change to a metric updates every report that uses it.
• How long does it take to get up and running? Connecting a data warehouse and defining an initial set of metrics can happen in a day or two. From there, it's ongoing work to expand the semantic model, add entities, and give Steep AI richer context.

Want to see it in action for your own data? Book a personalized demo with our team.