Datadog London Event Highlights AI-Powered Observability and LLM Monitoring Workshops

Datadog has published event materials for Datadog Live in London that place AI-powered observability and LLM monitoring at the center of its June 22, 2026 program.

Datadog frames observability around AI systems

According to Datadog’s event page for Datadog Live in London, the program includes coverage of AI-powered observability, LLM Observability, AI Guard, and Bits AI SRE. The event page also describes workshops focused on evaluating and optimizing AI agent performance using production traces.

The London agenda reflects a broader shift in enterprise monitoring: teams are not only tracking application uptime and infrastructure health, but also trying to understand how AI features behave in production. For applications built with large language models, that can include prompt and response behavior, latency, token usage, model performance, tool calls, and downstream application effects.

Hands-on LLM observability training

Datadog’s separate “LLM Observability with Gemini” workshop page describes a hands-on session in which participants analyze LLM traces, prompts and responses, evaluations, token usage, and latency. Datadog says the workshop also covers correlations with APM, logs, metrics, and infrastructure telemetry.

That framing suggests the company is positioning LLM observability as an extension of existing software monitoring rather than a standalone activity. In Datadog’s description, LLM traces can be examined alongside traditional operational data, which may help engineering teams investigate whether an AI feature’s issue stems from model behavior, application code, infrastructure, or a combination of factors.

Agent observability and production traces

Datadog’s Agent Observability product page says teams can build datasets from production traces, run experiments across prompts and models, inspect execution graphs and tool decisions, and correlate LLM spans with APM, infrastructure, and RUM sessions.

Those capabilities are aimed at AI applications that do more than return a single model response. Agentic applications may call tools, make intermediate decisions, and interact with external systems. Datadog’s product materials indicate that its observability approach is intended to show how those steps unfold and how they relate to the surrounding application environment.

The company’s product page also says teams can run experiments across prompts and models. That points to a use case beyond incident investigation: comparing changes before release or evaluating whether a prompt, model, or tool configuration affects performance and reliability.

AI assistance for operations teams

The London event page also references Bits AI SRE and AI Guard. Based on Datadog’s description, these offerings sit within its broader AI observability and operations theme. The provided event material does not give detailed technical specifications for each session, but it identifies them as part of the program’s focus on AI-powered observability.

For operations and platform teams, the relevance is straightforward: as AI functions are added to production software, monitoring needs to account for both conventional system behavior and AI-specific behavior. Datadog’s materials emphasize trace-level visibility, evaluation workflows, and correlation with existing telemetry sources.

Why it matters

Datadog’s London event materials show how observability vendors are adapting their products and training programs for LLM-based applications. The company is presenting LLM and agent observability as part of the same operational toolkit used for applications, logs, metrics, infrastructure, and user monitoring.

The sources do not establish how widely these tools are adopted or how they compare with competing products. What they do show is Datadog’s current emphasis: helping teams inspect AI application behavior in production, evaluate prompts and models, and connect LLM traces to the rest of the software stack.