Eclipse Opens Up Enterprise AI Agent Development with ADL (thenewstack.io)

The Eclipse Foundation has unveiled the Agent Definition Language (ADL) as the centerpiece of Eclipse LMOS (Language Models Operating System, now in alpha), an open-source effort to make enterprise AI agent development more accessible, reliable and scalable. ADL is a structured, model‑neutral language and visual toolkit that lets business domain experts define agent behavior as versionable standard operating procedures (not fragile prompts), enabling immediate testing and iteration without engineering ticket cycles. ADL grew from a JVM/Kotlin stack built at Deutsche Telekom to avoid rewriting enterprise APIs and DevOps — a pragmatic choice that has already powered multiagent, customer-facing systems at telecom scale. Eclipse LMOS comprises three integrated pieces: ADL (authoring and collaboration), the ARC Agent Framework (JVM-native Kotlin runtime with built-in debugging and rapid iteration tools), and the LMOS Platform (Cloud Native orchestration for lifecycle management, discovery, semantic routing and observability; built on CNCF/Kubernetes/Istio). The platform emphasizes auditability, versioning, multitenancy and reuse, addressing enterprise pain points like container sprawl and prompt engineering fragility. By aligning with existing JVM ecosystems and open governance, LMOS aims to be an open alternative to proprietary agent platforms — though it will contend with faster-moving standards (MCP, A2A) and hyperscaler ecosystems. The project targets “reliable AI” in operations and seeks community contributors to mature an enterprise-grade, vendor‑neutral agent stack.