Eclipse LMOS Introduces the Industry’s First Agent Definition Language (ADL) 

/ AI, Application Development, Open Source

Open-source innovation redefines enterprise-scale agentic AI with structured, model-agnostic governance

The News

The Eclipse Foundation has announced the addition of Agent Definition Language (ADL) to its Eclipse LMOS (Language Model Operating System) project, a major milestone in the evolution of agentic AI for enterprise environments.

ADL is the first open, model-neutral language for defining agent behavior and collaboration, addressing the growing complexity of prompt engineering and the lack of standardization in multi-agent systems. It provides a structured, versionable framework for business and engineering teams to co-define, test, and govern agent behavior, paving the way for enterprise-scale, maintainable AI orchestration.

The update reinforces Eclipse LMOS’s role as the open alternative to proprietary agentic AI platforms, allowing organizations to design intelligent multi-agent ecosystems using cloud-native, Kubernetes-based infrastructure that aligns with existing enterprise DevOps.

Analyst Take

Eclipse LMOS represents a shift in how AI agents are built, deployed, and managed. By introducing ADL, the project moves beyond freeform prompt engineering (long seen as the weakest link in enterprise AI scalability) toward structured, declarative agent programming.

ADL acts as a domain-specific language for AI agent design, allowing technical and non-technical stakeholders alike to collaboratively define agent roles, goals, and interaction rules. This not only improves consistency and auditability but also transforms agent design from an experimental art into an engineering discipline.

According to Mike Milinkovich, executive director of the Eclipse Foundation, “Agentic AI is redefining enterprise software, yet until now, there have been no open source alternatives to proprietary offerings.” With ADL, Eclipse provides enterprises with a transparent, extensible foundation for agentic systems, one that rivals closed platforms like OpenAI’s GPTs or Anthropic’s Claude Agents while offering greater control and sovereignty.

Enterprise-Grade Orchestration with Open Governance

Built on open standards such as Kubernetes and Istio, LMOS embodies the CNCF philosophy of composability and interoperability. The platform provides a cloud-native orchestration layer that manages agent lifecycles, routing, and observability while maintaining compatibility with existing JVM and Kotlin-based infrastructure.

This design allows enterprises to integrate agentic systems directly into existing workflows, CI/CD pipelines, and microservices architectures without replatforming. The addition of ADL further strengthens governance, offering version control, modular design, and schema validation for agent definitions.

Eclipse LMOS’s architecture spans three core components:

  • Eclipse LMOS ADL: The new model-agnostic language and visual toolkit for defining and editing agent behavior
  • Eclipse LMOS ARC Agent Framework: A JVM-native agent framework for developing and testing AI agents, with integrated visualization and debugging tools
  • Eclipse LMOS Platform: A vendor-neutral orchestration layer for multi-agent routing, lifecycle management, and monitoring

This modular approach mirrors what theCUBE Research has described as the “AI operating system era,” where modular, open systems replace siloed LLM wrappers and API integrations.

Real-World Validation with Deutsche Telekom’s Agentic AI

The Eclipse LMOS platform is already battle-tested through Deutsche Telekom’s Frag Magenta OneBOT assistant, one of Europe’s largest enterprise agentic deployments. Running millions of interactions across service and sales channels, this implementation validates LMOS’s resilience, scalability, and real-world applicability.

By processing millions of multilingual requests in production, LMOS demonstrates that open frameworks can meet (and even exceed) the reliability of commercial alternatives. This strengthens the Eclipse Foundation’s role as a key enabler of sovereign AI ecosystems, particularly in regions like the EU where open governance, transparency, and data compliance are strategic imperatives.

Collaboration Between Domain Experts and Engineers

A defining strength of ADL is its ability to bridge the gap between business users and developers. Instead of relying on prompt engineers to manually encode behavior, domain experts can now specify rules and expectations directly through ADL’s structured syntax or visual interface.

This approach shortens the iteration cycle between ideation and deployment, reducing time-to-value for enterprise AI initiatives. In doing so, LMOS effectively enables collaborative agent design, where business logic, compliance requirements, and technical architecture evolve in tandem.

As enterprises scale their use of multi-agent workflows (from customer service to autonomous IT operations), the need for shared governance languages like ADL will likely become a cornerstone of agentic system design.

Looking Ahead

The release of ADL establishes Eclipse LMOS as a leading open standard for agent definition and orchestration. By offering an extensible, vendor-neutral alternative to proprietary agent frameworks, the Eclipse Foundation is building the scaffolding for sovereign, interoperable agentic ecosystems.

In the coming year, expect LMOS to expand its developer ecosystem through additional SDKs, language bindings, and integrations with popular model providers. The move aligns with a broader industry shift toward open agent ecosystems, where interoperability and transparency drive enterprise adoption.

Ultimately, ADL’s structured approach to agent design could become as foundational to agentic AI as SQL was to databases, ushering in an era where agents are not just prompted, but programmed, governed, and scaled.

Key Takeaways

  • Eclipse LMOS introduces ADL, the world’s first open, structured language for defining agent behavior at enterprise scale.
  • Bridges business and technical collaboration, empowering non-engineers to shape AI agent logic.
  • Built for the enterprise, with Kubernetes-based orchestration, multi-tenancy, and JVM-native extensibility.
  • Validated at scale through Deutsche Telekom’s Frag Magenta OneBOT deployment.
  • Open and sovereign by design, establishing a transparent alternative to proprietary agent frameworks.

Authors

  • Paul Nashawaty

    Paul Nashawaty, Practice Leader and Lead Principal Analyst, specializes in application modernization across build, release and operations. With a wealth of expertise in digital transformation initiatives spanning front-end and back-end systems, he also possesses comprehensive knowledge of the underlying infrastructure ecosystem crucial for supporting modernization endeavors. With over 25 years of experience, Paul has a proven track record in implementing effective go-to-market strategies, including the identification of new market channels, the growth and cultivation of partner ecosystems, and the successful execution of strategic plans resulting in positive business outcomes for his clients.

    View all posts

  • With over 15 years of hands-on experience in operations roles across legal, financial, and technology sectors, Sam Weston brings deep expertise in the systems that power modern enterprises such as ERP, CRM, HCM, CX, and beyond. Her career has spanned the full spectrum of enterprise applications, from optimizing business processes and managing platforms to leading digital transformation initiatives.

    Sam has transitioned her expertise into the analyst arena, focusing on enterprise applications and the evolving role they play in business productivity and transformation. She provides independent insights that bridge technology capabilities with business outcomes, helping organizations and vendors alike navigate a changing enterprise software landscape.

    View all posts