Azure AI Foundry Observability
Microsoft AzureMicrosoft's GenAI observability surface inside Microsoft Foundry, providing evaluation, monitoring and tracing for LLM applications and agents. It is built on OpenTelemetry and feeds Azure Monitor Application Insights, with built-in evaluators for quality, RAG groundedness and relevance, safety, and agent-specific measures such as tool-call accuracy and task completion. Tracing covers LangChain, LangGraph, the OpenAI Agents SDK and the Microsoft Agent Framework.
Microsoft publishes no usage figure for this surface, so its adoption band leans on the reach of Microsoft Foundry rather than on a measurement of the surface. Verified 2026-08-13 via the Microsoft Learn observability concept page.
Openness
1 high confidence- license
- proprietary(Azure managed service)
- source
- closed
- note
- tracing built on open OTel standard + Azure Monitor App Insights, but the observability product is proprietary and Azure-locked
Proprietary Microsoft Azure managed offering. OTel is the wire standard (trace data is portable) but the service itself is closed.
- https://learn.microsoft.com/en-us/azure/foundry/concepts/observability recorded 2026-08-13
"Observability in generative AI" on Microsoft Learn. "Microsoft Foundry provides three core capabilities that work together to deliver comprehensive observability across the AI application lifecycle", the tracing one being "Built on OpenTelemetry standards and integrated with Azure Monitor Application Insights". OpenTelemetry is the wire format; the service consuming it is Azure's and is not published.
Adoption
3 low confidenceA feature surface within Microsoft Foundry on Azure; no standalone usage figures published for the observability capability specifically. Level 3 reflects availability inside a large cloud AI platform, not a measured user count; directional. The quantity actually banded is the reach of Microsoft Foundry rather than usage of its observability surface, and the cited page discloses no figure for either.
- https://learn.microsoft.com/en-us/azure/foundry/concepts/observability recorded 2026-08-13
The concept page, read for a usage figure and containing none - no customer count, no account count, no volume. It describes observability as three core capabilities of Microsoft Foundry, which places it inside the platform without saying how many people use it. No GA date appears on the page.
Capability
3 medium confidenceThree pillars - evaluation, monitoring and tracing - with strong OpenTelemetry and framework coverage: distributed tracing of LLM calls, tool invocations, agent decisions and cross-agent handoffs, capturing inputs, outputs, latency and token usage through Azure Monitor Application Insights, with auto-instrumentation across Semantic Kernel, LangChain, LangGraph and the OpenAI Agents SDK. Prompt management, datasets and annotation are not evidenced as first-class native features, so the matrix is thinner than the dedicated open source platforms and this sits a rung below them at 3.
- https://learn.microsoft.com/en-us/azure/foundry/concepts/observability recorded 2026-08-13
"Core observability capabilities - Microsoft Foundry provides three core capabilities", the first being Evaluation, whose "Evaluators measure the quality, safety, and reliability" of outputs, and tracing covering "agent decisions, and inter-service dependencies. Built on OpenTelemetry standards and integrated with Azure Monitor Application Insights". Quality gates are described as something you build into CI/CD rather than a product surface here.
Verified 2026-08-13