Microsoft Artificial Intelligence Mind Map

Version 3 of Microsoft AI Services classification organizes Microsoft’s AI portfolio along consumption model (how to access and run the service) and functional grouping.

High-level Classification

1. SaaS AI Offerings. Fully‑managed, turnkey AI embedded in business applications—no infra or code required.

  • Microsoft 365 features (Copilot, Search) surface generative and semantic capabilities directly in your productivity apps.
  • Dynamics 365 features (Copilot, Insights, Fraud Protection, etc.) embed AI into CRM/ERP workloads for sales, service, finance, supply chain and more.
  • Power Platform features (Copilot in Power Apps/Automate/BI, AI Builder, Virtual Agents) enable low‑code/no‑code makers to add AI into apps, flows and bots.
  • Microsoft Fabric features (Fabric Copilot, Data Science, Real‑Time Analytics, Vector Search) bring AI into the unified lakehouse and self‑service analytics experience.

2. PaaS AI Offerings. Platform services you build on—ranging from prebuilt AI APIs to full custom‑model lifecycles and hosting.

  • Azure AI Services (Cognitive APIs, OpenAI, Document Intelligence, Metrics Advisor, etc.) provide ready‑made Vision, Speech, Language, Search and Decision capabilities via REST.
  • Custom ML Models platforms (Azure ML, AI Foundry, Databricks, SynapseML, HDInsight) give you the tools to prepare data, train bespoke models and orchestrate experiments.
  • ML Model Deployment Services (AKS, ACI, Container Apps, Batch, ACR, ML Managed Endpoints) let you package those models as containers and serve them at scale or in batch.

3. IaaS & Edge AI Offerings. Raw compute and hybrid/edge runtimes for maximum control or offline scenarios.

  • GPU‑powered VMs and ML Compute Clusters for heavy training.
  • IoT Edge, Stack Edge and Cognitive Services Containers for on‑prem or disconnected inferencing.
  • Azure Arc to extend AI management across cloud, on‑prem and multicloud.

4. Authoring Tools. Developer and data‑science workbenches for building, experimenting and collaborating on models and AI apps.

  • Hosted notebooks (Azure ML, Synapse, Fabric) and Jupyter for interactive data exploration.
  • VS Code, PyCharm, Visual Studio, Codespaces + Copilot for code‑centric AI development.

Detailed Classification

SaaS AI Offerings

AI Features in Microsoft 365

  • Microsoft 365 Copilot. Tenant‑isolated LLM assistant embedded in Word, Excel, PowerPoint, Outlook and Teams.
  • Microsoft 365 Search. Semantic, Graph‑powered enterprise search across M365 (SharePoint, Teams, OneDrive, Outlook).

AI Features in Dynamics 365

  • Dynamics 365 Copilot (in multiple apps). Generative‑AI pane in Sales, Service, Finance, Supply Chain, etc., summarizing records and drafting text.
  • Dynamics 365 Customer Insights. CDP with AI‑driven identity resolution, churn & CLV prediction and dynamic segment discovery.
  • Dynamics 365 Business Central. SMB ERP with AI‑powered cash‑flow forecasting and inventory/sales predictions via Azure ML.
  • Dynamics 365 Sales Insights. Predictive lead/opportunity scoring plus conversation intelligence for calls and emails.
  • Dynamics 365 Customer Service Insights. AI‑based case classification, routing and knowledge suggestion to boost support metrics.
  • Dynamics 365 Supply Chain Management. Next‑gen demand forecasting and inventory optimization models running under the covers on Azure ML.
  • Dynamics 365 Finance Insights. Cash‑flow forecasting, anomaly detection and invoice payment prediction embedded in Dynamics Finance.
  • Dynamics 365 Field Service. Resource Scheduling Optimization and IoT‑driven predictive maintenance recommendations.
  • Dynamics 365 Fraud Protection. Adaptive AI risk scoring for payments, account creation and order protection based on global telemetry.

AI Features in Power Platform

  • Copilot in Power Apps/Automate/BI. NL prompts generate apps, flows and reports (DAX/narrative) in low‑code interfaces.
  • AI Builder. Low‑code ML model builder (prediction, form processing, object detection, text classification).
  • Power Virtual Agents. No‑code chatbot creator with integration to Dataverse and Azure AI Services.
  • Power Automate AI Features. Prebuilt AI actions (form processing, sentiment, key‑phrase, entity extraction) in flows.
  • Power BI AI Features. Built‑in AI visuals (Key Influencers, Anomaly Detection), Q&A and AutoML in Power BI.

AI Features in Microsoft Fabric

  • Fabric Copilot in Power BI. GPT‑4–powered assistant for report creation, DAX generation and narrative insights in Fabric lakehouse.
  • Fabric Data Science. Integrated AutoML and notebooks (Python/Scala) against OneLake data with MLflow tracking.
  • Fabric Real‑Time Analytics (KQL anomaly & forecasting). Native Kusto functions for anomaly detection and time‑series forecasting on streaming data.
  • Fabric Lakehouse Vector Search. First‑party vector store for retrieval‑augmented generation over lakehouse tables.

PaaS AI Offerings

Azure AI Services

  • Azure OpenAI Service. Hosted GPT‑family models (GPT‑3.5, GPT‑4) with enterprise security and compliance.
  • Vision APIs. Prebuilt Computer Vision and Face recognition services via REST APIs.
  • Speech APIs. Speech‑to‑Text, Text‑to‑Speech, Speech Translation and Custom Speech models.
  • Language APIs. Text Analytics, Translator, Language Studio/Authoring, QnA Maker for NLP tasks.
  • Custom APIs. Custom Vision and Custom Speech for training domain‑specific image/speech models.
  • AI Search (Cognitive Search). AI‑powered search index with semantic ranking and cognitive skill enrichment.
  • Azure Metrics Advisor. Time‑series anomaly detection and root‑cause analysis for monitoring applications.
  • Azure Immersive Reader. Reading assistance service that adds text‑to‑speech, translation and comprehension aids.
  • Azure Video Indexer. Video analysis for transcription, face/person detection, keyword extraction.
  • Azure AI Document Intelligence. Extract structured data (key/value, tables) from documents using prebuilt/custom models.
  • Azure AI Agent Service. Framework for building and hosting AI agents that combine LLMs, RAG and actions.
  • Bot Services (old service). Azure Bot Service & Composer for building conversational bots with SDK/tooling.
  • Content Safety. AI API for detecting harmful or unsafe content in text and images.
  • Content Moderator. Moderation APIs for text, images and videos to filter profanity, PII, adult content.
  • Personalizer. Reinforcement learning API for personalized content recommendations.

Custom ML Models

  • Azure Machine Learning Service. End‑to‑end ML platform for data prep, training, MLOps, deployment and monitoring.
  • Azure AI Foundry. Unified generative‑AI studio for LLM apps, prompt flows, RAG and AI agents.
  • Azure Databricks ML Runtime & MLflow. Managed Spark‑based ML environment with MLflow experiment tracking.
  • Azure Synapse Spark + SynapseML. Spark analytics and distributed ML inside Synapse with built‑in MLlib and SynapseML.
  • Azure Container Registry (ACR). Private Docker registry for storing and versioning container images.

ML Model Deployment Services

  • Azure Container Instances (ACI). Serverless containers for quick, per‑second billed inference.
  • Azure Kubernetes Service (AKS). Managed Kubernetes for scalable, production AI model hosting.
  • Azure Container Apps. Serverless micro‑container hosting with Dapr/KEDA support.
  • Azure Batch. Managed batch compute for large‑scale parallel inference/training.
  • Azure Machine Learning Managed Endpoints. One‑click real‑time/batch endpoints for models in Azure ML workspaces.
  • Azure ML Compute Clusters. Dedicated CPU/GPU clusters for distributed model training.

IaaS & Edge AI Offerings

  • Azure Virtual Machines (N‑series GPU). IaaS VMs with NVIDIA GPUs for custom training or inferencing workloads.
  • Azure Data Science VM. Preconfigured VM image with popular ML frameworks (legacy; use Azure ML compute instead).
  • Azure IoT Edge. Containerized deployment of AI modules on IoT devices for offline/edge inferencing.
  • Azure Stack Edge. On‑prem edge appliance with GPU/FPGA for local AI inferencing and preprocessing.
  • Cognitive Services Containers. Docker images of Cognitive Services APIs for on‑prem or air‑gapped environments.
  • Azure Arc. Hybrid management platform to deploy and manage AI services across on‑prem, edge and cloud.

Authoring Tools

  • Hosted notebooks in Azure ML or Synapse. Browser‑based Jupyter notebooks fully integrated into Azure ML or Synapse Studio for data science.
  • Jupyter Notebooks (in Azure ML, Synapse, Fabric). Open‑source notebook interface for interactive data exploration and model prototyping.
  • Visual Studio Code (Python, ML & Copilot extensions). Lightweight code editor with extensions for Python, ML, Docker and GitHub Copilot integration.
  • PyCharm. Python IDE with Azure plugins for local ML development and remote debugging.
  • GitHub Codespaces & Copilot. Cloud‑hosted development environments preconfigured with Copilot AI assistance.
  • Visual Studio. Full‑featured IDE with Azure ML and container tooling for building AI services in .NET.

Reference Materials

Unknown's avatar

Published by Ivan Kosyakov

Leave a comment