Our team worked the booth at Snowflake Summit last week. Hundreds of conversations across three days, and so many of them followed a pattern I’ve kept seeing.
Every booth on the floor is announcing the same thing. Every keynote promises the same agentic future. And that story is making everyone feel further behind instead of excited about what’s coming.
That fear is real, but it’s one we can get past. Because it isn’t the real problem.
What Snowflake Summit actually said
Let’s start with the announcements.
Snowflake and Anthropic deepened their partnership, putting Claude inside Snowflake Cortex AI so enterprises can run frontier reasoning on governed data without shipping that data somewhere else. Cortex Code, the coding agent now called CoCo, was named the fastest-growing product in Snowflake’s history. Horizon Catalog gained a new context layer, Horizon Context, built so that “every person, tool, and AI agent operates from the same trusted business context.” Snowflake Intelligence became CoWork, a personal agent for knowledge workers. The interoperable lakehouse story matured with Apache Iceberg v3: one live, governed copy of your data, queryable across engines without moving or duplicating it. CoCo picked up Datastream for real-time ingestion.
Christian Kleinerman, Snowflake’s EVP of Product, said it plainly: “As innovation accelerates, data fragmentation becomes the constraint.” Every headline at the event pointed to the same idea.
Trusted, governed, contextual data is the precondition for AI that works.
Why does everyone feel “behind”?
McKinsey put a number on it this year. In a 2026 survey of roughly 500 organizations, nearly two-thirds named security as the top barrier to scaling AI. The thing slowing enterprise AI adoption down is the foundation underneath it: governance, trust, context. The unglamorous layer nobody wants to demo on the keynote stage.
With new tools announced every day, it feels like everyone else is years ahead of you and pulling away. They aren’t. A foundational data gap means everyone is standing in the same mud, and the company that clears it first is the one that gets out first.
The industry is focused on the wrong part
Archimedes supposedly said, “Give me a lever long enough and a place to stand, and I will move the world.”
AI is definitely a lever. It is a genuinely long one. But right now everyone is bragging about the lever, not about what they’ve moved with it.
The real work is finally starting to show up, at least in the conversations. Deciding the one thing you want to move is the first step. Then you need a fulcrum, and good, consented data is what gives the lever something to push against.
Otherwise you’re just holding a very long stick.
Clarity about outcomes was missing from most of the booths. It was starting to show up in the conversations. I hope that continues.
We had our own announcements at Snowflake Summit
Our team brought Tealium Audience Discovery to the Summit this year. It is a Snowflake Native App, running inside your own Snowflake account, on your roles and your compute. A marketer can describe an audience in plain language through Snowflake Cortex, an analyst can write it in SQL with an assist, and the result is validated with live row counts and previews before anything is sent to your connectors and channels.
This app is the lever. It turns the customer data you already own into revenue. Your data is the strongest asset you have, and we have been saying that for years. Now there is a real lever in your toolbox to put that foundation to work.
What to do today
If you came home from Summit, or are just reading about it, feeling like there’s a clock ticking down, do these things instead of panicking.
Name an outcome you want to move. Hint: it’s not “adopt AI.” It’s something you could measure by Friday: a churn cohort, a re-engagement window, the minutes between a customer’s intent and your response.
Build on what you already have. Kleinerman is right that fragmentation is the constraint, so if you’ve invested in Snowflake as your data cloud, build on the governed data where it already sits.
Get the data right before a model touches it. Fresh, consented, unified, real-time customer data. If your model is running on stale, fragmented data, it will produce stale, fragmented results, and it will do it so confidently you’ll believe it.
Measure the impact. Run small tests of change. Learn what it takes to get the outcomes you want, because that is what helps your team get out of the mud and actually feel the benefits of AI.
You’re not behind. You’re right where everyone else is.
Stop reading AI as a deadline you are failing to meet. It is a lever, already long enough, lying on the ground in front of you. The only question that matters is the one no keynote will answer for you. What do you want to move?
