Skip navigation
More AI use ≠ more impact.
0:00 / 0:59

More AI use ≠ more impact.

sridhar

@RamaswmySridhar47K followers

Visit on X
202 likesDownloadOriginal post

56,048 views

More AI use ≠ more impact. Today, Snowflake is introducing dynamic model routing in Cortex AI Gateway + support for DeepSeek-V4-Flash 0731 and GLM-5.3. A step toward intelligence efficiency: turning AI resources into tangible business impact. More in my latest blog 👇

10 Comments

Comments from the original post on X (Twitter)

  • Tarun Tyagi's profile picture

    @thetaruntyagi

    @Snowflake More AI use is not more impact, agreed. Impact shows up when the output triggers an action in a system of record instead of a dashboard. That gap is our whole focus: #AgenticAI #AgenticAutomation #AIAutomation

  • The Hive's profile picture

    @theHiveryIQ

    @Snowflake Routing well is one problem. Proving which model answered is another. When one model grades another, the router is its own auditor. We filed on binding model identity into the inference receipt, so substitution breaks the commitment.

  • Dirt Jordan's profile picture

    @techfusionjb

    @Snowflake Dynamic routing is where real savings live. I stopped buying the 'more compute = more impact' pitch years ago.

  • MaeveKnows's profile picture

    @maeveknows

    @Snowflake The efficiency is very much needed

  • Ankur Kumar's profile picture

    @ankurkumarz

    @Snowflake Dynamic model routing is the need of the hour and that’s why from coding agents to Search and Agent platforms need to support that leveraging cost-effective models first than high reasoning, high effort frontier models 👍🏻

  • AI Mastery Guide's profile picture

    @aiseomastery

    @Snowflake Efficiency over raw usage, agreed

  • the_translator's profile picture

    @HereForRatio2

    @Snowflake What a dork?

  • Manoj Chandra Jha's profile picture

    @ManojChandraJha

    @Snowflake Here is our POV from

  • Eric Kang's profile picture

    @Eric_Kangg

    @Snowflake Dynamic routing becomes useful when every task has a visible owner and acceptance check — not just a model picker. Same pattern with DeepSeek workers.

  • johndemic's profile picture

    @johndemic

    @Snowflake Routing is becoming table stakes. The harder part is preserving an audit trail of why a model was selected, what it cost, and whether the workload achieved its outcome. Does Cortex expose that decision trail?