The leading cloud monitoring and security platform has launched a new suite of observability tools designed to empower organizations deploying AI agents and LLM-powered applications. Announced at DASH 2025, Datadog’s annual observability conference, these features offer full-stack monitoring, structured testing, and security insights into the growing complexity of generative AI systems.
Bridging the Gap in AI Agent Observability
As generative AI and autonomous agents move from concept to production, most organizations struggle with a lack of visibility into how their agents perform, interact with tools, and impact real business goals. Datadog’s new offerings AI Agent Monitoring, LLM Experiments, and AI Agents Console—close this visibility gap.
These features are part of Datadog’s LLM Observability product, designed to help teams monitor agent workflows, evaluate prompt and model changes, and govern third-party agent usage—all from a centralized interface.
Key Features to Monitor, Test, and Secure AI Systems
AI Agent Monitoring
This tool visualizes the full decision path of any AI agent—from inputs and tool calls to final outputs—via an interactive graph. It lets developers investigate latency spikes, misfired tool actions, and looping behaviors while correlating them with cost, performance, and security metrics.
“Agents are not just chatbots they’re autonomous systems. And for them to be production-ready, observability must be as advanced as the agents themselves,” said Timothée Lacroix, CTO of Mistral AI. “Datadog’s solution enables just that.”
LLM Experiments
Teams can now test changes to prompts, models, or app logic using real production data or synthetic datasets. By running structured experiments, users can measure response accuracy, throughput, and costs, making it easier to prevent regressions and validate improvements.
This tool is especially relevant for teams pushing Claude 4 or other advanced LLMs into real-world applications—from customer support to R&D.
Driving Business Impact from AI Investments
Datadog’s VP of Product Yrieix Garnier pointed to a recent study showing that only 25% of AI initiatives deliver measurable ROI. “Our goal is to change that,” Garnier said. “With these new tools, we’re making it easier for companies to justify AI budgets and build production-grade systems that work.”
By uniting observability with AI development, Datadog is helping enterprises and startups accelerate innovation while keeping their systems robust, efficient, and compliant.
Conclusion
Datadog’s LLM Observability suite represents a major leap forward in how organizations manage and optimize agentic AI systems. As the complexity and responsibility of AI agents grow, tools like these are becoming essential for teams building reliable, scalable, and high-impact AI solutions.
With monitoring, experimentation, and governance now unified, Datadog is setting a new standard for AI observability in production.
About The Author
