AI is everywhere, and at Oracle, we’re not just adding AI—we’re architecting it into our technologies. As AI, data, and infrastructure converge, the boundaries between your database, models, compute, and orchestration are disappearing, giving developers a more unified way to build, train, and deploy intelligent applications. 

Here’s how that convergence is showing up across Oracle’s latest AI innovations: 

AI for Data 

The newly released Oracle AI Database 26ai brings AI directly into the database and enables developers to automate workflows and build agentic AI capabilities. The release includes: 

  • Select AI Agent – lets you define, run, and manage AI agents inside the database via REST or MCP servers. 

  • Unified Hybrid Vector Search – allows developers to query vectors alongside relational, JSON, graph, and spatial data in one step. When paired with LLMs and MCP, it enables agentic workflows that fetch additional context and produce richer, more accurate results with your own data inside the database. 

  • Flexible Model Choices – choose from ONNX embedding models, integrate with leading LLM providers, or run private inference through Oracle’s Private AI Services Container to help keep data secure. 

Autonomous AI Lakehouse was also just released, providing an open, interoperable data platform to accelerate how data teams build AI and analytics solutions. With native support for Apache Iceberg, teams can use open-source tools and popular languages to access their Iceberg data on any platform—operational or analytic—in the cloud or on-premises.  

By unifying relational, text, JSON, knowledge graph, and spatial data, Oracle AI Database simplifies how developers work with complex information and enables AI tools and agents to reason more intelligently over your data. Try AI Database at FreeSQL.com, and for dev/test, try Oracle AI Database Free.  

Open Agent Specification (Agent Spec) 

With Agent Spec, it’s easier to work across different agent frameworks and scale across platforms. As a declarative, framework-agnostic standard for defining AI agents and workflows, it lets you describe what an agent should do in a unified format. Agents and workflows can be defined once, covering tasks, tools, control flow, and memory, and executed across different runtimes without rewriting logic, which helps improve portability, validation, and reusability. It also enables tighter integration with Oracle’s AI stack, allowing agents built with Select AI or MCP-enabled workflows to connect more seamlessly with other tools or runtime environments. Try it out. 

Multicloud Universal Credits 

Oracle’s new Multicloud Universal Credits gives developers the flexibility to deploy and manage Oracle AI Database workloads across multiple clouds, offering portability, cost control, and the freedom to build wherever it makes the most sense. Learn more 

Together, these innovations show what’s possible when AI, data, and infrastructure truly converge, giving developers the power to build, deploy, and scale intelligent applications faster than ever, using familiar languages, open formats, and multicloud flexibility. Stop by booth 400 at KubeCon to see AI innovation in action!