Databricks + Google Cloud

Databricks and Google Cloud are expanding their alliance around a simple pitch: let companies build artificial intelligence agents on their own data without loosening control over who can see or use it. (databricks.com) Databricks said April 16 that the partnership now centers on Gemini integrations, Google Cloud Marketplace distribution, and “governed AI” built on Unity Catalog, its system for managing access to data and AI assets. The company said the two firms have more than 2,500 joint customers and have worked together since 2021. (databricks.com) The underlying problem is familiar to large companies: generative artificial intelligence systems are more useful when they can reach internal documents, databases, and tools, but that also raises the risk of exposing sensitive information or producing answers from the wrong source. Databricks describes Unity Catalog as a unified governance layer for data and AI assets across its platform. (databricks.com) Google and Databricks first formalized the Gemini tie-up on June 12, 2025, when Databricks said Google’s models would become native products inside its Data Intelligence Platform. The companies said that setup would let customers build and scale AI agents on enterprise data inside Databricks rather than moving workloads across separate systems. (databricks.com) That pitch has shifted from chatbots to “agents,” shorthand for software that can search, summarize, call tools, and complete multi-step tasks with limited human prompting. In the Databricks roadmap, those agents are supposed to run against governed company data, with Gemini handling model inference and Databricks handling data, permissions, and orchestration. (databricks.com) Google Cloud is making the same case outside classic analytics. Avid and Google Cloud said April 16 they signed a multi-year partnership to embed Gemini models and Vertex AI into Avid Media Composer and Avid Content Core for film and television post-production. (avid.com) In that workflow, the “agent” is less a chatbot than an assistant inside editing software: Avid said the tools are meant to automate media discovery, metadata generation, clip search, and other labor-intensive tasks in large video libraries. The companies said they plan to demonstrate the new workflows at the NAB Show in Las Vegas, scheduled for April 18-22, 2026. (avid.com) Google Cloud Next 2026, where Databricks is showcasing the partnership, is set for April 22-24 at Mandalay Bay in Las Vegas. The timing puts both announcements in the same week, with Google Cloud pressing a broader message that its models should sit inside industry-specific software, not just in standalone chat interfaces. (googlecloudevents.com) What these deals do not settle is how much autonomy companies will actually allow these systems in production, or how often humans will stay in the loop for approvals and edits. For now, both announcements are framed around enterprise control: the agent can move faster, but the rules, data boundaries, and customer workflows stay in place. (databricks.com)