Dynatrace

platform

proprietary-SaaS(Grail-powered, Davis AI)

license

commercial

ingestion

OpenTelemetry+OpenLLMetry(via Traceloop)+GenAI-semconv(open standards, not open product)

source

closed

Dynatrace AI Observability is a commercial module of the proprietary Dynatrace platform. It ingests data via OpenTelemetry and OpenLLMetry, open standards consumed by a closed product.

• https://docs.dynatrace.com/docs/license recorded 2026-09-29

  License page: "Dynatrace Platform Subscription (DPS) is a consumption-based licensing model designed for flexibility." so the product is sold under a commercial subscription. Searched the body for open source: none.

• https://docs.dynatrace.com/docs/observe/dynatrace-for-ai-observability recorded 2026-09-29

  Docs page: "Monitor and debug Agent interactions such as function calling, LLM calls, tool-use, RAG, and resolve performance, latency, cost, and reliability issues." Dynatrace scores quality with "online LLM-as-a-judge evaluations" and offers the open-source dt-evals CLI; the page describes no self-hosted or on-premises Dynatrace distribution and no published Dynatrace source code.

Dynatrace publishes no usage figures specific to AI Observability, and the documentation confirms only that the feature exists, with no customer, account or ingestion-volume count. Rather than infer adoption from Dynatrace as a platform, the map leaves this product unscored.

• https://docs.dynatrace.com/docs/observe/dynatrace-for-ai-observability recorded 2026-09-29

  The capability documentation, read for a usage figure and containing none: it says the feature is used "to monitor performance (token consumption, latency, availability, and errors) at scale" and never states a customer, account or user count. Searched the body for customers, adoption, million and billion: none.

Dynatrace pairs end-to-end tracing and cost tracking with online LLM-as-a-judge evaluation of production traces, through the open-source dt-evals CLI with more than 10 built-in judges, custom judges, CI quality gates and drift tracking. It documents no prompt versioning, datasets or prompt playground, which Langfuse bundles with its evaluation, so the two are level.

• https://docs.dynatrace.com/docs/observe/dynatrace-for-ai-observability recorded 2026-09-29

  Documentation for the capability - scoring is described as "Score the quality, safety, and relevance of GenAI and agent responses with online LLM-as-a-judge evaluations", the open-source dt-evals CLI "includes more than 10 built-in LLM judge evaluators", tracing covers "Trace prompt flows from initial request to final response", and Cost is "Token usage, service fees, or overall resource consumption".

• https://docs.dynatrace.com/docs/observe/dynatrace-for-ai-observability/get-started/llm-judge-evaluation recorded 2026-09-29

  Tutorial page - covers "How to add a custom LLM-as-a-judge evaluator alongside the built-in catalog" and "How to automate recurring runs and gate a CI pipeline on quality", with results pivoting from any low score to the source trace.