Transform 2025: Why observability is critical for AI agent ecosystems

The era of autonomous software is rapidly approaching, as highlighted at Transform 2025, where industry leaders Ashan Willy, CEO of New Relic, and Sam Witteveen, CEO and co-founder of Red Dragon AI, discussed the significance of observability in AI-driven systems. They emphasized the importance of establishing measurable returns on investment (ROI) while navigating the evolving infrastructure landscape. New Relic has positioned itself as a leader in observability, capturing and correlating real-time telemetry data from applications, logs, and infrastructure. This approach goes beyond mere monitoring; it equips teams with the necessary context and insights to comprehend, troubleshoot, and optimize complex systems, even when faced with unforeseen challenges. The integration of generative and agentic AI has added layers of complexity to these tasks. The demand for advanced observability tools has surged, with New Relic reporting a 30% quarter-over-quarter increase in the adoption of its AI monitoring capabilities. Willy noted a remarkable diversity in AI models being utilized by enterprises, with a staggering 92% increase in the variety of models since the introduction of foundational models like GPT. This shift raises critical questions about measuring effectiveness across different use cases, which vary significantly by industry and organizational goals. Observability’s focus has evolved since New Relic’s inception in 2008, transitioning from application monitoring for SaaS and mobile environments to addressing the needs of cloud infrastructure. The surge of AI and agentic AI has shifted attention back to applications where micro-agents and nano-agents are actively generating AI-written code. As the complexity of services and microservices grows, the challenge of managing observability increases, often overwhelming human operators. Willy asserts that AI can alleviate these burdens by enabling cooperative modes of operation, where agents automate routine tasks and enhance the visibility of observability across multiple tools within an organization’s ecosystem. This integration allows developers to receive immediate alerts about code errors, facilitating quick resolutions without leaving their coding environments. For instance, if an issue arises with code in GitHub, an observability platform can detect the problem, propose solutions, and even automate the resolution process. The latest announcement from New Relic includes a collaboration with Copilot, a coding assistant that identifies specific lines of code causing issues, rectifies them, and prepares a new version for deployment. As companies embrace agentic AI, Willy emphasizes that understanding the mechanisms of observability will be vital to leveraging its full potential. He concludes, "As you start to build all these agentic integrations and pieces, you’re going to want to know what the agent does, and this reasoning is essential for understanding what’s happening in your production environment." Observability is poised to play a pivotal role in the future of AI ecosystems, ensuring organizations can harness the capabilities of their agentic technologies effectively.