Snowflake enters the agentic AI race with a desktop platform that turns governed enterprise data into finished business deliverables. The 9,100 AI accounts it already has will be the first test of whether data gravity translates to workflow adoption.

Snowflake announced Project SnowWork on March 18, 2026, an autonomous AI platform designed to help business users get work done faster. The idea is simple: a business user describes what they need in plain language, and Project SnowWork plans the steps, pulls from governed Snowflake data, runs the analysis, and delivers a finished output. That could be a board-ready forecast deck, a spreadsheet flagging churn risks, or a report identifying supply chain bottlenecks. It is launching in research preview to a limited set of customers.

In an interview with Axios, CEO Sridhar Ramaswamy walked through a practical example: a user could ask SnowWork to build a pitch deck, and the platform would pull data from multiple sources, organize it, and draft an accompanying email, all without writing any code. The company also disclosed that 9,100 accounts are now using its AI products as of the end of February 2026, a useful baseline for measuring how quickly SnowWork adoption grows from here.

What It Does

Project SnowWork comes with pre-built skill profiles for common business roles in finance, sales, marketing, and operations. These profiles understand the workflows, terminology, and key metrics each role cares about. The platform handles multi-step tasks end to end, from querying data to generating structured deliverables with recommended next steps, all within a single interaction. It inherits Snowflake's existing role-based access controls, data masking policies, and audit logging, so the AI operates within the same security boundaries as the data it touches.

"We're moving past the 'chatbot' phase and into an era where AI doesn't just tell you what's wrong, it helps you fix it." Bala Kasiviswanathan, VP of Developer & AI Experiences, Snowflake

That framing matters because it draws a clear line between SnowWork and the wave of conversational AI assistants that can answer questions but stop short of executing work. Project SnowWork joins two other AI products in Snowflake's lineup. Snowflake Intelligence, now generally available, is the company's enterprise knowledge agent for natural language question-answering. Cortex Code is an AI coding agent for data engineering and application development. Project SnowWork fills the third lane: workflow execution for non-technical business users who need results, not raw data access.

Why It Matters

Most business users today still depend on data teams to answer basic questions. They file a request, wait days, and get a static report. AI assistants can help with general tasks, but they lack access to an organization's governed data, shared metric definitions, and security controls. Project SnowWork is Snowflake's answer to that gap. Because it runs on top of the same platform that already stores and governs the enterprise data, the outputs are grounded in trusted numbers rather than general-purpose AI guesswork.

Ramaswamy framed this as Snowflake's move toward the "agentic enterprise," where AI is not just answering questions but driving decisions and completing work. That framing is shared across the industry right now. Salesforce, Microsoft, ServiceNow, and Google are all building agentic capabilities into their platforms. Snowflake has not disclosed pricing for SnowWork, a timeline for broader availability, or which organizations are in the initial preview cohort.

What to Watch

Snowflake is not the only data platform vendor moving in this direction. Databricks launched Genie Code on March 11, an agent that handles tasks like building pipelines and debugging failures. The pattern is clear: every major data platform is racing to add an execution layer on top of its governed data assets. SnowWork alone does not hand Snowflake a lasting competitive edge over other data platforms with similar ambitions.

The more interesting question is whether data platform vendors as a category will deliver more trustworthy AI execution than standalone large language models that lack secure access to proprietary enterprise data. On that front, the governed data advantage is real.

For technology leaders, the evaluation question is practical. If your organization already runs significant workloads on Snowflake, Project SnowWork is worth tracking as it moves beyond research preview. If not, the broader competitive landscape still has room to develop. The agentic enterprise is coming. The vendors whose agents consistently deliver governed, accurate, useful results will be the ones that earn long-term trust.

About Shashi Bellamkonda

Loading bio...

LinkedIn