Cybersecurity

Snowflake's security pitch leaves work undone

Snowflake says partners will secure 150,000 agents, but names no proofs on identity sprawl, cost or response.

StartupHub.ai Staff

By StartupHub.ai Staff, reviewed by Daniel Singer, Editor

2 min read

Security operations view for AI agents at scale
Credit: Snowflake

Snowflake says its partner network is ready to secure the agentic enterprise.

I do not buy ready. The post from Radhika Janardanan and Nitika Gupta makes a fair claim, that no single vendor can cover identity, data protection, access and response alone, then stretches it to suggest coverage across seven fronts is already here. The scale argument is real, with Snowflake citing Gartner that the average global Fortune 500 will run more than 150,000 agents while only 13% of organizations say they have the right governance in place.

Agents are not normal apps.

Snowflake gets that part right. A traditional application runs predefined code, while agents make decisions, call tools and move data without a human in the loop, so Snowflake offers Horizon Catalog governance, a distinct identity for every agent and policies to protect sensitive data. That foundation pitch follows a week of Snowflake telling every team to center on governed data, including sales. The problem is what would have to be true for partners to finish the job.

Tags would have to follow sensitive data into every tool, cost centers and reporting lines would have to drive access without manual work, and 150,000 agent identities would have to be reviewed and retired on time. Snowflake names the categories, from posture management and tokenization to runtime protection and recovery, but names no vendors in the text, no integration depth, no latency or cost figures and no detection precision. Longer retention for security telemetry in Snowflake is the strongest promise here, because fragmented telemetry does slow investigations.

It still assumes an orderly lifecycle. Cyera warns that identities, credentials, workloads and permissions can appear and disappear inside a single workflow as agents generate code, invoke tools and delegate to other agents. That ephemeral sprawl is harder than provision, review and decommission.

If you are building, assume the catalog is the start, not the control, and test deprovisioning and abuse detection before you scale to thousands of agents.

Sources

Get the free daily brief every morning

Every morning the day's AI news, new AI startups and new funding rounds. Free and takes a few minutes to read, unsubscribe at any time.

Join 64,000 other subscribers.

TopicsSnowflakeSnowflake Horizon CatalogGartnerCyeraRadhika JanardananNitika Gupta

StartupHub.ai Staff

Written by

StartupHub.ai StaffEditorial team

The staff writers of StartupHub.ai, ranging from investment analysts to avid AI tool users, early adopters and critical enthusiasts. Backgrounds span engineering, business and the arts. We hold every piece to rigorous standards of research and review.

Daniel Singer

Reviewed by

Daniel Singer is an AI expert who has been writing about, covering and building AI technology since 2018.

Test your Agent Readiness now

Paste your website. How well ChatGPT, Claude and the other agents can read and use a site, in about a minute.

Watch

The videos behind the news, one flip at a time

The feed mixes the videos from our articles with startup demos and you flip through them like TV channels for free with no signup.