From Attendee to Sponsor: What We Learned at Databricks Data + AI Amsterdam

Three thousand data practitioners packed into RAI Amsterdam this week for the Databricks Data + AI World Tour, and for the first time, Cosmos Thrace wasn’t just in the audience – we were on the floor as an Explorer sponsor with our own booth.

The difference? Instead of taking notes from the back of the room, we were having direct conversations about what keeps data leaders up at night.

The Real Conversations Happening at Booth 12.19

The polished keynotes gave us the “what’s next” in data and AI. But our booth gave us something more valuable: unfiltered insight into what companies are actually struggling with right now.

The pattern was clear. Company after company arrived with the same fundamental challenge: they know Databricks is where they need to be, but they’re stuck in legacy systems that feel impossible to migrate without disrupting operations.

One conversation stands out. A data engineering lead from a major European retailer spent 20 minutes walking us through their current architecture – a patchwork of on-premise warehouses, cloud data lakes, and reporting tools that barely talk to each other. “We know the lakehouse model makes sense,” he said. “We just don’t know how to get there without breaking everything.”

That’s the gap we’re built to bridge.

Three Technical Shifts Worth Watching

While we were heads-down at the booth, three keynote sessions highlighted where Databricks is pushing the platform:

1. Unity Catalog is becoming non-negotiable Governance used to be the thing you’d tackle “eventually.” Now it’s table stakes. Unity Catalog’s new capabilities around access control, lineage tracking, and multi-cloud management mean enterprises can finally govern data and AI workloads without creating bottlenecks. If you’re planning a Databricks implementation in 2025, Unity Catalog needs to be in scope from day one.
2. Agent Bricks tackles the 80% AI failure rate Most AI projects fail before production because teams can’t guarantee accuracy at scale. Agent Bricks uses synthetic data generation and automated benchmarking to optimize AI agents for both cost and quality before deployment. This matters because it removes the guesswork from building multi-agent systems that need to operate reliably in high-stakes environments.
3. Streaming data now flows directly into governed tables Confluent’s Tableflow integration with Unity Catalog eliminates the custom pipeline engineering that traditionally sits between Kafka streams and Delta tables. Real-time data can now materialize as governed, analytics-ready tables automatically, cutting both complexity and total cost of ownership by up to 60%.

What Databricks Data + AI Amsterdam Did for Us

Brand positioning: We’re no longer the boutique consultancy that “also does Databricks.” We’re now visibly part of the Databricks partner ecosystem, standing alongside the platform’s most active implementation partners.

Pipeline development: We walked away with qualified conversations – not vague “let’s stay in touch” exchanges, but specific migration challenges with timelines and budgets attached.

Validation: The problems we’ve been solving for clients aren’t edge cases. Migration complexity, governance uncertainty, and AI production readiness are universal pain points. Our specialization matters.

Next Stop: Databricks Data + AI Stockholm

We’re doing this again on December 10 at the Stockholm leg of the Data + AI World Tour. If you’re attending and want to talk about Databricks migrations, Unity Catalog rollouts, or getting AI agents into production, find us there.

And if you couldn’t make Amsterdam but have questions about navigating your own Databricks journey, whether that’s platform migration, lakehouse architecture, or AI implementation—reach out. We’re here for exactly these conversations.