Building data platforms in the AI era

Most data tools are pitching themselves as "AI-native" right now. This conversation goes the other way. The platforms holding up in the AI era were built on solid architecture before AI workflows ever mattered — declarative modeling, semantic clarity, governed metrics. The same properties that make data legible to humans make it legible to LLMs. AI didn't define that standard. It validated it. Siavoush Mohammadi (Co-founder & CPO, Daana) has spent sixteen years building data platforms across companies of every size. Johan Baltzar (Founder & CEO, Steep) built a metrics-first BI platform that's reshaping how organizations consume their data. Together, they dig into where the lines sit between platform, semantic layer, and consumer — and why those lines matter more as AI becomes the primary consumer. What you'll take away: What "AI-era" actually means for data platforms, and what the "AI-native" framing gets wrong Why semantic clarity is the unlock that lets an LLM reason reliably about a business (with a real Nextory example) How declarative modeling and a governed metrics layer make self-service real, not theoretical What's still hard, what's solved, and what's overhyped Who this is for: Data leaders, CTOs, and data engineers and architects thinking about how to build or rebuild a data platform that holds up as AI moves from experiment to expectation. Steep is a metrics-first BI platform built for how modern data teams actually work. Learn more at steep.app