Welcome to AI Country

If you’re driving north on Highway 101 into San Francisco, you’ll see a new message before you hit the city limits:

Welcome to AI Country.

It’s a simple statement, but it carries a few layers. First, it’s a nod to what San Francisco has become—the epicenter of AI innovation. Every other billboard in this stretch of highway now has “AI” somewhere on it. It’s the new gold rush. 

As one of the first billboards on the route from SFO into the city, we just wanted to say, "Welcome."

But it’s also a reminder: beneath all the hype, only a small percentage of AI projects actually succeed. 

MIT researchers recently found that just 5% of AI initiatives deliver measurable business value. Look closer at that 5%, and they share something in common: a focus on internal, operational workflows. Automation that improves how work gets done day-to-day, not just another chatbot demo or flashy prototype.

That’s exactly where Sigma lives.

We’re not an AI lab or a model-hosting platform. Snowflake and Databricks already do that exceptionally well by providing secure, scalable, and enterprise-grade foundations for data and AI. 

But those platforms aren’t where most people work. They’re the engine room. Sigma is the interface, the UI to enterprise data and AI.

What I love about Sigma is how fast you can go from looking at data to building something useful. Whether it’s approvals, planning models, or forecasts—it all happens live in the same place: the data warehouse. Sigma’s UI ensures anyone can work with data in a familiar way, and because it’s tied directly to the warehouse, everything stays secure, auditable, and up to date.

And now, Sigma is a 100% AI-driven platform.

That shows up in three ways:

1. All of Sigma is powered by agents: Sigma now runs on top of hundreds of agents. These agents execute extremely specific tasks like adding a chart, formatting text, and creating pivot tables.
   
   Anything you build in Sigma uses agents. And Sigma leverages another set of agents that orchestrate these task agents.
   
   
2. AI Builder. Sigma’s newest UI that lets anyone build apps and workflows with AI assistance. This lowers the barrier for who gets to build—not just who gets to analyze.
   ‍
3. ‍AI Query and MCP: Apply LLMs and other 3rd-party agents in spreadsheet-like functions. MCP extends this further by connecting Sigma to an emerging ecosystem of interoperable AI and data services—enabling organizations to tap into shared models, custom compute, and domain-specific agents that work seamlessly within Sigma’s workflow.

Unlike other prompt-to-app tools that generate unsecured prototypes and require separate deployment checks, AI apps in Sigma inherit your warehouse’s security by design. That means the n+1 app can scale across your org without extra approval cycles or manual reviews.

It’s how enterprises move from experimenting with AI to operationalizing it—exactly what the 5% of successful AI projects have figured out.

For Sigma, the billboard is a small marker of a bigger shift. The conversation around AI is moving past the novelty of models and prompts and into the hard, necessary work of putting it to use inside real businesses.