Reflections from the Data Innovation Summit 2025
A while ago, I attended the Data Innovation Summit in Sweden, a two-day annual summit where the future of data takes center stage. Unsurprisingly, the buzz around AI was everywhere. Where previous editions focused on data engineering, BI, and visualizations, this year it was all about preparing for an AI-driven future.
Let me share my key insights.
“AI will widen the competitive gap, but only if it’s built on correct and governed data”
A quote that really stuck with me. Or in other words: despite all the tools and models available, without high-quality and well-governed data, you’re not going to get far. Many of the session emphasized this foundation:
- Data governance: from metadata management and lineage to policies and ownership.
- Data quality: tools like Sifflet, Soda, and Great Expectations enabling automated validation across the data stack.
- Shift to the left: moving data quality and governance responsibilities earlier in the pipeline, with more ownership and investment at the business level.
- Data as a product: domain-oriented ownership models (think data mesh) are gaining serious traction.
These pillars are no longer nice-to-haves. They’re prerequisites.
Put the customer first, not the technology
One key theme that emerged throughout the summit was this: Start from the problem, not the solution.
Before jumping into AI, organizations need to stop and ask: What are we really trying to solve? Is AI even the right solution? Is the business aligned? Are we ready from a data and capability perspective?
Several talks stressed the importance of organizational readiness: Do you have executive sponsorship? Do your teams understand the tech? Can you handle unstructured data? Is ownership embedded where it needs to be?
These aren’t technical questions, they’re strategic ones. And they reflect a broader mindset: one that puts the customer’s problem before the technology. Because at the end of the day, data only creates value when it’s actually used, by real people, solving real problems.
From POCs to impact: what it really takes
Another takeaway was that many companies are still experimenting, lots of proof-of-concepts, few scalable rollouts. And that’s okay. But to move beyond that stage, you need more than ambition. You need change: cultural, structural, and human. That means involving business units early, creating internal champions, embracing failure as part of the process, and embedding feedback loops in your workflows. AI is not an IT project. It’s a company-wide journey. And don’t forget: this field is moving incredibly fast. What is innovative today might be irrelevant next year.
That’s why flexibility and a solid foundation matter more than ever: governance, quality, shared ownership, long-term thinking and getting people on board.
Or simply, what we call, having a togaether mindset.
