Last week, I joined thousands of technologists, engineers, and data visionaries at Snowflake Summit 2025 DEV DAY in San Francisco. The energy was palpable. From early morning keynotes to hands-on labs and late-day networking, it was clear that Snowflake is no longer just a data warehouse; it’s evolving into a full-blown development and AI platform.

As someone who’s constantly exploring new technologies and evaluating what’s worth bringing into our workflow at Glatco Inc., I didn’t attend with a sales pitch in mind. I came with questions:

  • How far has Snowflake come as a development platform?
  • Is it just hype, or are real-world teams building real things?
  • Where does AI fit into all this?

Here’s what stood out:

1. Native Applications Are Coming to the Data Cloud

Session: Building Resilient, Scalable Native Apps with Snowpark Container Services
Speaker: Charles Yorek, Snowflake

Charles Yorek walked us through how Snowflake’s new Snowpark Container Services lets developers deploy full containerized applications within the Snowflake environment itself. This is a major shift—rather than pulling data into external compute, you can now bring your app logic to the data.

He shared examples where microservices and third-party ML libraries run as containerized workloads directly inside Snowflake, reducing latency and eliminating complex orchestration across separate platforms. The service supports Kubernetes-like autoscaling, integrated IAM, and tight coupling with Snowflake data policies. It opens the door to running production-grade logic as part of the data pipeline itself.

One thing that really stood out to me in Charles Yorek’s session was how Snowpark Container Services protects developers’ intellectual property; SQL queries executed inside your container aren’t exposed in Snowflake’s logs. As someone who’s built and deployed backend services, this felt like a big deal. You’re not just moving compute closer to the data; you’re embedding secure, production-ready app logic right into the platform, without giving up visibility or control over your runtime. That kind of balance between integration and isolation is rare, and it opens up real architectural flexibility.

2. Agentic Apps in Snowflake: Powered by Cortex and MCP

Session: Building Agentic Applications in Snowflake
Speakers: Arun Agarwal & Dash Desai, Snowflake

In this session, we explored how to build a Data Agent using Snowflake Cortex AI, packaged in a Streamlit application, that can intelligently respond to natural language questions by reasoning over both structured and unstructured data. The dataset used was intentionally artificial, providing a clean and controlled environment free from external model knowledge, ideal for testing and evaluation.

Key components included:

  • Cortex Analyst for semantic modeling and querying structured data.
  • Cortex Search for handling unstructured data such as PDFs and images.
  • Integration of these tools through the Cortex Agents REST API, enabling unified data intelligence.

By the end, participants, including me, had a working AI-powered Data Agent embedded in Streamlit, capable of understanding natural language questions, retrieving and reasoning over both structured and unstructured data, all while running securely within the Snowflake environment and providing a clean, user-friendly interface.

It was fun getting hands-on with something that usually feels theoretical. I left with a clearer understanding of where Snowflake is heading with AI and how much easier it’s getting to build secure, intelligent workflows natively in the data cloud.

Refined Takeaway: Making Data Work for You, Intelligently

Instead of lofty predictions about the future of data, here’s what hit home during our day at the summit:

Snowflake is making it easier to turn data into action. Whether it’s Snowpipe Streaming simplifying ingestion, Container Services embedding secure logic at the data layer, or Cortex enabling smarter automation through agents—these tools are about practical gains, not just promises.

And personally? As someone who’s constantly balancing delivery speed with long-term maintainability, I appreciated seeing how Snowflake is folding modern capabilities right into the platform. No bolt-ons, no handoffs, just a more unified way to build.

Final Thoughts

At Glatco, we’re always exploring new ways to build smarter, simpler systems. Snowflake’s latest platform extensions, while still maturing, offer promising tools for teams looking to streamline workflows, reduce data movement, and explore practical AI integrations. We’ll continue to evaluate how and where they might fit in our own roadmap.

For now, it’s encouraging to see infrastructure, data, and intelligence coming together in more accessible ways.

Let’s Connect and Explore Together

At Glatco, we’re passionate about helping businesses unlock the full potential of their data—efficiently and intelligently. Reach out to us to start a conversation about how modern data platforms and AI-driven workflows can power your next breakthrough.

Get in touch with our team.