Artificial intelligence is evolving at an incredible pace, and open source innovation is keeping stride. Across the Eclipse Foundation, an expanding ecosystem of projects is redefining what AI can do, with efforts that span developer tools, data governance, trustworthy model design, and intelligent agent orchestration. What ties them all together is a shared belief that AI should be open, transparent, and built for the real world.
One of the most exciting projects in this space, Eclipse LMOS (Language Model Operating System), just took a big step forward with the introduction of Agent Definition Language (ADL), a new way for teams to design, deploy, and manage AI agents and multi-agent systems at scale.
Moving from prompt engineering to true agent design
We have all seen how powerful AI can be, but getting consistent results from language models can still feel like trial and error. Traditional prompt engineering is often a mix of art and luck, and it does not always hold up when you need to scale or govern AI behavior.
That is where ADL comes in. It offers a structured, model-agnostic framework for defining agent behavior that teams can version, test, and improve collaboratively. Instead of relying on scattered prompts or undocumented tweaks, ADL gives you a shared language for how agents think and act so business experts and engineers can finally work together in a unified way.
The result is faster development, stronger alignment between teams, and AI systems that are easier to monitor, adjust, and trust over time.
Built for the enterprise you already have
What makes LMOS so practical is how seamlessly it integrates with existing enterprise environments. It is a cloud native, vendor-neutral platform that runs on familiar technologies such as Kubernetes, Istio, and JVM-based systems. That means you can build and scale advanced multi-agent systems without replacing your infrastructure, retraining your teams, or reinventing your processes.
In short, LMOS meets you where you are. It allows enterprises to leverage their existing people, tools, and workflows while unlocking powerful new AI capabilities.
That is not just theoretical. At Deutsche Telekom, LMOS powers one of Europe’s largest enterprise deployments of agentic AI. Their Frag Magenta OneBOT assistant, built on LMOS, has handled millions of customer service and sales interactions across multiple countries, demonstrating the platform’s scalability, resilience, and enterprise readiness.
Open source as a strategic advantage
There is no shortage of AI platforms out there, but most are proprietary systems that limit flexibility and transparency. LMOS takes a different path. As an open source project, it is built for collaboration, auditability, and long-term sustainability.
By choosing LMOS, enterprises gain:
- Cross-team collaboration between business and technical stakeholders
- Scalable agent orchestration across diverse, hybrid environments
- Transparent governance and freedom from vendor lock-in
- Multi-tenant support to meet complex enterprise requirements
In a landscape dominated by black box solutions, LMOS offers a truly open and trustworthy alternative that organisations can understand, extend, and build on together.
Join the community
Eclipse LMOS is just one piece of a much bigger story unfolding at the Eclipse Foundation. Across our ecosystem, open source communities are advancing AI in all kinds of ways, from agent orchestration and MLOps to trustworthy, data-driven design.
If you are building something exciting in the AI space, we would love to hear from you. Whether you want to contribute to an existing project or bring your own initiative to the Foundation, there is a place for you in this growing community.
Visit eclipse.org/ai to learn more, share your ideas, and get involved.
