← Field Guide  ·  Tier 2 · Hyperscaler & Vendor References (Cross-Validation)

Microsoft

Agent Factory · Azure AI Foundry · Semantic Kernel

Hyperscaler · cloud doc7 patterns
In Plain English

This page shows every pattern that Microsoft has published or written about — 7 in total. Each one includes a simple picture of how it works, a short explanation, an example of when to use it, and its trade-offs (the good parts and the not-so-good parts).

Reading patterns from one source at a time matters because it lets you see how a single company or team thinks about building AI agents, and compare their ideas side by side. If a pattern here also appears in the bigger Deep Catalogue, you will see a link so you can read the full, detailed version.

An open reference book with a pencil in the fold under warm lamp light

2A.01Tool-Using Agent (Azure)

AZURE TOOL-USING AGENT Agent OpenAPI Spec Validator Tool

Agent invokes declared tools with validated parameters and structured returns — Azure AI Foundry implementation of typed tool use.

Use caseEnterprise agents on Azure calling internal APIs, data stores, partner services.
Trade-offsPortable via OpenAPI / MCP · Azure-flavoured tooling.

See 1C.02 Tool Use

2A.02Reflection Agent (Azure)

AZURE REFLECTION Agent Draft Reviewer Revisedraftreviewrevise

Azure AI Foundry reference for generator + reviewer loop. Maps to evaluator-optimiser.

Use caseHigh-stakes generation (compliance drafts, code).
Trade-offsQuality lift · Doubles cost.

See 1A.07, 1C.01

2A.03Planning Agent (Azure)

AZURE PLANNER Planner Worker 1 Worker 2 Synthesiser

Azure reference for Plan-and-Execute with planner model and workers.

Use caseStructured multi-step tasks on Azure.
Trade-offsCheaper than ReAct on predictable tasks · Less adaptive.

See 1C.03, 1D.06

2A.04Multi-Agent Collaboration (Azure)

AZURE MULTI-AGENT Orchestrator Agent A Agent B Agent C

Azure reference for specialist teams with explicit handoff protocols.

Use caseComplex enterprise workflows spanning multiple domains.
Trade-offsScales capability · Coordination overhead.

See 1C.04

2A.05Semantic Kernel Planner

SEMANTIC KERNEL PLANNER Goal Plannerpicksfunctions Plan Function Calls Result

Microsoft's Semantic Kernel planner dynamically composes registered functions (Sequential, Stepwise, Handlebars variants).

Use caseAgents built on the Semantic Kernel runtime.
Trade-offsFramework-native · Tied to SK ecosystem.

2A.06Magentic-One

MAGENTIC-ONE Orchestrator Web Surfer File Surfer Coder Terminal

Microsoft Research multi-agent orchestrator: orchestrator + four specialists (web-surfer, file-surfer, coder, terminal).

Use caseBrowser-based and computer-use tasks spanning web + files + code.
Trade-offsStrong generalist · Heavyweight.

2A.07Agent Factory Observability

AGENT FACTORY OBSERVABILITY Agent Run Span LLM Call Spans Tool Call Spans Guard Spans

Azure-native pattern for span-level tracing of agent execution integrated with App Insights.

Use caseProduction agents requiring enterprise-grade observability.
Trade-offsDeep Azure integration · Vendor-coupled.

See Deep Catalogue OBS-ACT-01 →

Also in Tier 2
AWS 4Salesforce 4OpenAI 3IBM & Others 2