Cloud Renames Contribute
Deprecated → MigratedAI & AgentsAzure

Azure Metrics Advisor Microsoft Fabric Real-Time Intelligence & Azure OpenAI (Cognitive Services Retirements)

Legacy names Azure Metrics Advisor, Azure AI Anomaly Detector, Azure Personalizer

Time-series anomaly detection, operational metrics monitoring, and reinforcement-learning personalization workloads migrating from retired Cognitive Services Decision APIs to Microsoft Fabric Real-Time Intelligence, Azure Monitor KQL, and Azure OpenAI.

Naming timeline

  1. Cognitive Services Anomaly Detector & Personalizer

    · Microsoft Build 2019 Preview (GA 2020)

    Launched under Azure Cognitive Services Decision APIs for time-series anomaly detection and contextual bandit reinforcement learning.

  2. Azure Metrics Advisor (Applied AI Services)

    · Microsoft Ignite 2020 Preview (GA June 2021)

    Built on Anomaly Detector to provide an AIOps and root-cause diagnostic workspace under Azure Applied AI Services.

  3. Microsoft Fabric Real-Time Intelligence & Azure OpenAI (Cognitive Services Retirements)Current name

    · Deprecation Notice (Retired May–Oct 2026)

    New resource creation blocked September 20, 2023; Metrics Advisor retired May 18, 2026, Personalizer retired August 25, 2026, and Anomaly Detector retired October 1, 2026 in favor of Microsoft Fabric Real-Time Intelligence, Azure Monitor KQL, and open-source anomaly-detector / learning-loop.

What replaced Azure Metrics Advisor?

Azure Metrics Advisor has been superseded by Microsoft Fabric Real-Time Intelligence & Azure OpenAI (Cognitive Services Retirements), with the latest change on 20 September 2023 (Deprecation Notice (Retired May–Oct 2026)). Earlier names covered by this entry also include Azure AI Anomaly Detector, Azure Personalizer. Time-series anomaly detection, operational metrics monitoring, and reinforcement-learning personalization workloads migrating from retired Cognitive Services Decision APIs to Microsoft Fabric Real-Time Intelligence, Azure Monitor KQL, and Azure OpenAI. The underlying legacy API or CLI surface is deprecated, so a technical migration is required.

Also known as

  • Azure AI Metrics Advisor
  • Azure Applied AI Metrics Advisor
  • Cognitive Services Anomaly Detector
  • Azure AI Personalizer

Technical identifiers

Legacy identifiers deprecated. The underlying API or CLI surface for the legacy name is officially deprecated, so this change requires a technical migration rather than a label update.

ARM Resource Providers (Microsoft.*)Microsoft.Fabric/capacities Microsoft.OperationalInsights/workspaces Microsoft.CognitiveServices/accounts (kind: AnomalyDetector - Deprecated) Microsoft.CognitiveServices/accounts (kind: MetricsAdvisor - Deprecated) Microsoft.CognitiveServices/accounts (kind: Personalizer - Deprecated)
Azure CLI (az) / PowerShellaz fabric az monitor log-analytics az cognitiveservices account
Entra ID / Azure RBAC RolesFabric Administrator Monitoring Contributor Cognitive Services Metrics Advisor Administrator Cognitive Services User
Terraform Resources (azurerm_*)azurerm_fabric_capacity azurerm_log_analytics_workspace azurerm_cognitive_account

The Microsoft.CognitiveServices/accounts kinds 'AnomalyDetector', 'MetricsAdvisor', and 'Personalizer' are retired as of May–October 2026; migrate time-series anomaly workloads to Microsoft Fabric (Microsoft.Fabric/capacities) or Azure Monitor Log Analytics KQL.

Notes & disambiguation

  • On September 20, 2023, Microsoft announced the deprecation of three Cognitive Services Decision APIs—Azure Metrics Advisor, Azure AI Anomaly Detector, and Azure Personalizer—immediately blocking the creation of new resources for Anomaly Detector and Personalizer from September 20, 2023.
  • Azure AI Metrics Advisor was retired on May 18, 2026, Azure Personalizer retired on August 25, 2026, and Azure AI Anomaly Detector retired on October 1, 2026 (at the end of the September 30, 2026 three-year notice window).
  • Official Microsoft migration guidance directs time-series monitoring and multivariate/univariate anomaly detection workloads to Microsoft Fabric Real-Time Intelligence (KQL series_decompose_anomalies() and the open-source microsoft/anomaly-detector library in Fabric Data Science) and Azure Monitor KQL anomaly detection, while contextual reinforcement learning workloads migrate to the open-source microsoft/learning-loop and Azure Machine Learning / Azure OpenAI.

Sources & official documentation

Catalog last updated . Spotted an error? Suggest an edit