News

Spring AI 2.0 Goes GA with Vector Search, Memory, and Agents on Azure Cosmos DB

Microsoft and the Spring team announce Spring AI 2.0 GA, bringing vector search, memory management, and agent support to Azure Cosmos DB. Plus: Spring Boot 4.1 features, CVE surge, and self-correcting structured output.

July 4, 2026· 2 min read· Source: This Week in Spring - June 30th, 2026
Spring AI 2.0 Goes GA with Vector Search, Memory, and Agents on Azure Cosmos DB

Spring AI 2.0 has reached general availability, and Microsoft's Cosmos DB team is first out of the gate with a deep integration post. The release brings first-class support for vector search, conversational memory, and AI agents on top of Azure Cosmos DB. If you're building LLM-powered apps in the Spring ecosystem, this is the upgrade you've been waiting for.

Spring AI 2.0: What's New

The headline features are vector search (pgvector-style, but for Cosmos DB), persistent memory for multi-turn conversations, and a structured agent framework. Craig Walls also published a recipe for hooking up ElevenLabs voice synthesis with Spring AI — useful for voice-enabled chatbots or accessibility features.

Christian Tsolov's blog on self-correcting structured output in Spring AI 2.0 is worth a read. It shows how the framework can automatically retry and fix malformed LLM responses, which is a common pain point in production.

Spring Boot 4.1 and the CVE Surge

Spring Boot 4.1 has been out for a few weeks now, shipping with auto-configuration for MongoDB-backed Spring Batch repositories — a nice quality-of-life improvement for batch workloads. But the bigger story is the CVE volume. Josh Long notes that AI-assisted vulnerability discovery has driven CVE reports from one or two per month to dozens in recent weeks. The team is patching and releasing fixes to Maven Central, but the disclosure cadence means you should update before the advisory page lights up.

Spring Boot 3.5.16 is also available as a patch release for those still on the 3.x line.

Agent Skills with Spring AI

Baeldung published a tutorial on building agent skills with Spring AI, covering how to define tool-using agents that can call external APIs or databases. This is a practical entry point for teams looking to move from simple RAG to agentic workflows.

Community and Ecosystem

InfoQ's Spring roundup covers the broader ecosystem news. CodeZine (Japanese) has a new Spring Boot series. And the latest Bootiful Podcast episode with DaShaun Carter dives into the CVE situation and what teams should be doing.