Is it just me, or is every data conference at the Javits Center this year the exact same conference? A couple of weeks ago it was MongoDB’s turn, and today I just got back from the Snowflake World Tour. It felt like deja vu. The lights are different, the logos are different, but the theme being blasted from the keynote stage is identical: AI is here, and we’re the platform you need for it.

Snowflake’s version of this story is called Cortex Agents. This was the star of the show, where they pitched a future of building and deploying AI that lives and runs right on top of your data. The vision is compelling, for sure. They even tried a live demo, which promptly ran into some real-world chaos in the form of bright red “Internal server error” messages. It was a good reminder that this stuff is still on the bleeding edge, even for the folks building it.

Snowflake World Tour Keynote

One session that stood out from the main AI theme was on Snowflake Postgres. It was an interesting deviation, and after their acquisition of Crunchy Data, it’s a clear indicator of where they’re heading. They’re making a run at developers building transactional apps, trying to pull OLTP workloads onto their platform and stop being just the final destination for data.

Of course, you can’t unleash AI agents on corporate data without locking them down, which is where the session on the Horizon Catalog provided a necessary dose of reality at the end of the day. They covered the governance and security guardrails, which is the stuff that actually matters if you want to use any of this in a real company.

So, walking out of the Javits Center again, the deja vu was real. The industry is shouting one thing right now: AI. Every platform wants to be the center of this new world. Last time it was MongoDB pitching itself as the “memory” for agents. Today, Snowflake’s vision was to be the entire “central nervous system”—the secure, governed platform for everything. It feels like the definition of a data platform is changing in real-time. For engineers, the job is the same as always: separate the signal from the noise.