Snowflake's CoCo: When Data Infrastructure Becomes the Agentic Layer
Cloud Platform EngineeringAt Snowflake Summit 2026, the company’s narrative shifted from “AI features in your data platform” to “your data platform is the agentic layer.” CoCo is the product that makes that argument concrete.
What Happened
Snowflake announced CoCo (Coding Agent), a native agent capable of orchestrating data workflows without a human writing SQL or managing pipeline logic. CoCo is available as a standalone desktop app and integrated into VS Code and Microsoft Excel. From any of those surfaces, users can automate workflows, build apps, and operationalize AI on enterprise data through natural language prompts.
The positioning is deliberate: Snowflake’s leadership described CoCo and its companion CoWork as “truly joyous products to use in terms of getting things done,” framing the experience as fundamentally different from building with a chat interface. The focus on joy is a product strategy claim — the company believes interface quality, not raw capability, determines enterprise adoption.
Why It Matters
Enterprise data platforms have historically been infrastructure: powerful, reliable, and deeply unsexy to use. The workflow was always the same — a data engineer writes the query, a platform team manages the pipeline, an analyst reads the output. CoCo breaks that model by making the agent the interface to the data, not a consultant or SQL practitioner.
The VS Code and Excel integration choices are significant. VS Code is where engineers live. Excel is where the business lives. An agent that can work in both surfaces — and can push outputs from Snowflake into the spreadsheet open on a finance analyst’s screen — changes who can interact with enterprise data infrastructure without changing the underlying data architecture.
This is consistent with the broader pattern in agentic AI infrastructure for enterprise: platforms that can serve as the data substrate for agentic workflows have a structural advantage over point solutions that need to integrate with a separate data layer.
What This Means for Builders
Teams building on Snowflake should evaluate CoCo’s orchestration capabilities for data pipeline automation tasks that currently require custom code or dbt. The Excel integration specifically opens a path to reduce the gap between data infrastructure and business-facing outputs — a persistent pain point in enterprise data teams.
For platform teams evaluating data warehouse vendors, Snowflake’s bet is clear: the agentic interface to your data is part of the platform value, not a layer above it. That changes the vendor evaluation criteria from “best query performance” to “best agent-native experience for our business users.”
The So What
Snowflake is making the case that enterprise data infrastructure and agentic AI are converging, and that the winner is whoever builds the most usable agent experience on top of the most reliable data layer. “Joyous” is a product north star, not a marketing word — it’s telling you that Snowflake thinks the current standard for enterprise AI tooling is joyless. They’re probably right, and that’s the gap they’re building toward.
Source: SiliconAngle
Content created with AI assistance and reviewed for accuracy.
Join the conversation
Stack Insiders is our free community for readers who want to go deeper — share resources, ask questions, and connect with others across every vertical we cover.
Join Stack Insiders →