Two big data-platform companies posted AI changes in their release notes this week: Databricks on 8 and 9 October, Snowflake on 7 and 9 October. All are previews or betas, and neither company gives a price or benchmark in these notes.
Databricks: Genie One can run code
Genie One is Databricks’ name for its AI experience for business users [1]. Confirmed (Databricks release notes), Public Preview. On 9 October the notes say: “Genie One can now run code in a secure, isolated sandbox to take on work that goes beyond SQL, such as more advanced analyses.” A sandbox is a walled-off space where code cannot touch the rest of a system. The notes add that the code runs “using your existing permissions, so Unity Catalog governance and data access controls continue to apply” [1]. Unity Catalog is Databricks’ layer for controlling who can see which data.
On 8 October, in Beta, “A Genie Agent can now search and take actions in external tools like Google Drive and Microsoft 365 through MCP connectors” [1]. MCP, the Model Context Protocol, is a standard way for AI tools to plug into other software; see our explainer. Databricks had already made similar connectors generally available for Genie One and Genie Code on 10 September [1].
The page warns: “Releases are staged. Your Databricks account might not be updated until a week or more after the initial release date.” [1]
Snowflake: agent-led marketplace search, and a decision model
Confirmed (Snowflake release notes), public preview, 9 October. Agentic Marketplace Discovery lets users “switch between Agentic discovery and Marketplace search”. You “Describe what you need to CoCo to find and compare matching listings, or search for providers and listings directly” [2]. From the conversation you can explore a listing, install it, start a trial or send a sales request, where the listing and your role allow it [2].
Confirmed, private preview, 7 October. Snowflake Decision is “a model available through AI_COMPLETE that evaluates text and structured data against user-defined questions”, returning “structured answers with probabilities” for choices, rubric scores and true-or-false statements [3]. AI_COMPLETE is the Snowflake function the model is called through. It is the same idea as the decision models TSN explained earlier this week.
What this does not show
- Prices. None of these notes states one. Separately, Databricks says Genie One and Genie Agents usage by users is free through 31 January 2027, but does not say whether code execution is covered [1].
- Performance. There are no benchmarks for any of these features.
- Wide availability. Previews and betas are not general releases, and Databricks’ rollout is staged.
- Independent testing. TSN read the notes only.
The Bottom Line
Databricks has put code execution for Genie One into public preview and extended its Genie Agents to Google Drive and Microsoft 365 in beta. Snowflake has begun a public preview of agent-led marketplace search and a private preview of a decision model. These are vendor statements about early features, not proof they work well.
Related on TSN: OpenAI’s Decisions API and Perplexity’s Open Decider: AI That Answers With a Probability, Not an Essay; MCP Explained: The USB-C for AI Tools (Model Context Protocol Guide)
Sources
- Databricks, “AI/BI and Genie One release notes 2026”, Databricks on Google Cloud documentation (company release notes; last updated 9 October 2026; entries of 9 October, 8 October, 10 September and 15 July). https://docs.databricks.com/gcp/en/ai-bi/release-notes/2026
- Snowflake, “Oct 9, 2026: Agentic Marketplace Discovery (Public preview)”, Snowflake release notes (company documentation). https://docs.snowflake.com/en/release-notes/2026/other/2026-10-09-agentic-marketplace-discovery-preview
- Snowflake, “Oct 7, 2026: Snowflake Decision with AI_COMPLETE (Private Preview)”, Snowflake release notes (company documentation). https://docs.snowflake.com/en/release-notes/2026/other/2026-10-07-ai-complete-snowflake-decision-preview
