
What happened
GitHub released the Copilot SDK for Java (version 1.0.7-preview.1), a client library that lets Java developers build AI agents without depending on a specific framework or AI vendor. The SDK works with any model provider—OpenAI, Azure, Anthropic, or custom endpoints—via BYOK (bring-your-own-key) support and requires no Copilot subscription to use with third-party models.
Why it matters
Until now, Java developers had to choose between Langchain4j (which tied them to that library) or Spring AI (which tied them to Spring's design choices). The new SDK is the first truly framework-agnostic approach for enterprise Java, supporting both Jakarta EE and Spring, and uses familiar Java patterns (CompletableFuture, annotations, lambdas, virtual threads) so existing Java teams can adopt it without learning a new paradigm.
What to watch
The SDK is available as a Maven dependency and requires JDK 17 or later (JDK 25 recommended for virtual threads). The annotation-based tool API (@CopilotTool) is marked experimental and requires passing the compiler flag -Acopilot.experimental.allowed=true. A complete sample app (real-estate lead-management agent) is available on GitHub running on Open Liberty 26.0.0.5 with Jakarta EE 11.
Summaries like this, in your inbox every morning.
The release addresses a real pain point in Java enterprise development: until now, building AI-driven applications required developers to accept a dependency on a middleware library (Langchain4j) or a specific framework ecosystem (Spring AI). The GitHub Copilot SDK for Java breaks that lock-in by providing a framework-agnostic, vendor-neutral foundation. The BYOK (bring-your-own-key) support means teams can swap AI providers without rewriting their application code, and the lack of a mandatory Copilot subscription requirement signals that this is positioned as a general-purpose agent orchestration library, not just a Copilot-consumption tool.
The SDK's design philosophy aligns with the article's stated belief in open standards: it uses familiar Java primitives (annotations, CompletableFuture, CDI beans, virtual threads) rather than forcing developers into a new model or API shape. The sample application—a Jakarta EE 11 real-estate agent using Open Liberty 26.0.0.5—demonstrates production-grade integration patterns, including proper use of managed thread factories to respect container lifecycle and context propagation. The experimental status of the @CopilotTool annotation API suggests the core agent session and tool registration mechanism is considered stable, but the declarative tool definition surface is still subject to change.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Airbnb CTO Ahmad Al-Dahle said about half of support tickets are now resolved purely by AI, 60% of code is AI-…

At Fortune's AIQ Summit, Booking Holdings CFO Ewout Steenbergen said hyperscalers spending hundreds of billion…

PwC Japan's spring 2026 six-country survey found only 9% of Japanese companies got effects that "greatly excee…

Gartner predicts over 40% of agentic AI projects will be canceled by the end of 2027

GitHub published career advice from Gwen Davis saying developers must direct AI agents, evaluate their output…

Google's September roundup is headlined by Gemini 4 Argon, a frontier model with a 1-million-token context win…
