Making AI Real for Business, but Data Still Comes First
by Nicolas Damus, Data Engineering Consultant
After spending several days at Snowflake Summit 2026, one thing became impossible to ignore: we’re well and truly entering the era of AI agents.
This year’s theme, Making AI Real for Business, was reflected in almost every keynote, breakout session, workshop and product demonstration I attended.
Whether the conversation was about analytics, customer engagement, governance or application development, it seemed to eventually come back to the same destination: AI agents.
Snowflake is clearly betting big on a future where agents become a standard part of how businesses interact with data.
While the event showcased significant innovation across AI, governance, data sharing and real-time analytics, it also highlighted an important reality: successful AI outcomes still depend on strong data foundations.
AI Agents Were The Main Character
This year, AI agents took centre stage.
Snowflake presented a future where agents don’t just answer questions; they automate workflows, retrieve information, extract insights, make recommendations, interact with enterprise systems and assist users across a range of business functions.
Instead of users manually navigating dashboards, querying databases or orchestrating workflows, intelligent agents can increasingly perform these tasks on their behalf.
New capabilities such as CoWork, Cortex Sense and Semantic Studio are designed to make AI more accessible while helping organisations operationalise AI at scale.
The vision is compelling. The technology itself is genuinely exciting.
For many organisations, this can significantly shift how business users engage with data and technology, and it’s not difficult to imagine the productivity gains organisations could achieve by putting capable agents in front of their data.
But after hearing similar messages repeated across multiple presentations and workshops, I found myself wanting more discussion about what happens underneath the agent.
What this means for organisations
For organisations exploring AI, the conversation is rapidly shifting towards agents that can interact with systems, execute workflows and automate decision-making processes.
Organisations should focus on where they can deliver measurable value and ensure the right data and governance foundations are in place.
Open Data Architectures Continue to Matter
While the AI announcements grabbed most of the headlines, I found some of the supporting platform investments equally interesting.
The ongoing support for Apache Iceberg and Polaris shows Snowflake’s recognition that organisations want flexibility and openness within their data ecosystems. Snowflake’s commitment enables organisations to work across diverse data ecosystems without becoming locked into a single platform.
As enterprises increasingly adopt hybrid and multi-cloud strategies, open architectures are becoming critical for ensuring flexibility, portability and long-term scalability.
This reflects a broader industry trend: organisations want the freedom to choose the best tools for their workloads while maintaining a unified approach to governance and data management.
What this means for organisations
Open and interoperable data architectures give organisations greater flexibility as their technology landscape evolves.
They make it easier to adopt new tools, integrate additional data sources and avoid vendor lock-in, helping businesses respond faster to changing requirements.
Governance Takes Centre Stage
Similarly, the expansion of Horizon Catalog and Horizon Context reflects a growing understanding that governance isn’t optional in an AI-first world.
Snowflake introduced enhancements through Horizon Catalog and Horizon Context, providing organisations with greater visibility and confidence into data assets, lineage, access controls, metadata and accountability.
The same applies to security and agent identity, another topic that featured prominently throughout the event.
In many ways, these announcements felt less flashy than the AI demonstrations, but arguably more important.
Strong governance is essential not only for regulatory compliance but also for ensuring confidence in AI-generated outputs.
The emphasis on governance throughout the Summit suggests Snowflake understands that enterprise AI requires more than powerful models. It requires transparency, accountability and control.
Why this is important
The success of enterprise AI depends on trust. Organisations need clear visibility into their data, who owns it and how it’s governed to ensure AI delivers reliable outcomes.
Real-Time Data and Streaming Continue to Evolve
Another theme I noticed was the growing focus on real-time capabilities.
Announcements around managed Kafka services, Datastream and Adaptive Compute continue Snowflake’s push towards supporting more event-driven and operational workloads.
This matters because AI agents are only truly useful when they have access to relevant and current information.
As organisations move beyond simple chat interfaces and begin building agents that support operational decision-making, the ability to process and respond to data in real time becomes increasingly valuable.
As AI increasingly relies on real-time context, investments in streaming infrastructure will likely become just as important as advances in model capabilities.
Why this is important
Organisations investing in real-time data pipelines and event-driven architectures will be better positioned to deliver responsive customer experiences, automate operational decisions and maximise the effectiveness of AI agents for real-time intelligence.
Security and Agent Identity Become Critical
With AI agents taking on more responsibility within enterprise environments, Snowflake also placed significant emphasis on security, observability and agent identity.
This reflects an emerging challenge across the industry.
As autonomous systems gain access to data and business processes, organisations need new mechanisms to understand what agents are doing, what they can access and how they are making decisions.
The future of enterprise AI will depend not only on agent capabilities but also on organisations’ ability to monitor, govern and secure them effectively.
Why this is important
As AI agents gain access to sensitive systems and business processes, organisations need to think beyond traditional user security models.
Understanding what agents can access, what actions they can perform and how those actions are monitored will become a critical part of enterprise risk management.
Security strategies will need to evolve alongside AI adoption.
AI Agents Are Powerful, but You Can’t Prompt Your Way Around Data Problems
One narrative I noticed throughout the conference was the suggestion that many business challenges could be solved by introducing an AI agent into an existing process.
In theory, that sounds appealing.
Need better customer insights? Ask an agent.
Need faster reporting? Ask an agent.
Need operational efficiencies? Ask an agent.
The reality is often more complex.
As data practitioners, we see what happens when underlying data quality, governance or integration issues are ignored.
AI agents don’t remove those challenges. If anything, they amplify them.
An agent is only as good as the data it can access.
If the data is fragmented across systems, poorly governed or difficult to trust, the agent will simply deliver faster answers with the same underlying problems.
Reliable AI agents depend on high-quality, well-governed and accessible data. Without strong data engineering practices, robust governance frameworks and trusted data foundations, even the most sophisticated AI solutions will struggle to deliver consistent business value.
That’s why one of my biggest takeaways from Summit wasn’t actually about AI.
It was about data foundations.
As organisations accelerate their AI initiatives, the businesses that achieve meaningful outcomes will likely be those that invest as much in their data platforms, governance and operational foundations as they do in AI itself.
Final Thoughts
Snowflake Summit 2026 demonstrated that enterprise AI is rapidly moving from concept to reality.
The direction is clear.
Snowflake sees AI agents becoming a fundamental layer between people and data, and they’re investing heavily to make that vision a reality.
I don’t doubt that agents will become an important part of how organisations work.
What I do question is whether some businesses will underestimate the amount of foundational work required to get there.
The organisations that succeed with AI won’t necessarily be the ones deploying the most agents. They’ll be the ones with the strongest data foundations, governance frameworks and engineering practices underpinning those agents.
That’s the part that can’t be solved with a prompt.
AI may have dominated the conversation at Summit 2026, but data remains the thing doing the heavy lifting.
