Visual Framework

Interactive decision trees to guide Microsoft AI technology selection.

Use these diagrams after working through the Decision Framework and Evaluation Criteria. They reinforce the nine critical questions and are designed to support workshops, architecture reviews, and executive readouts.

Diagram Index
Complete Decision Flow
Persona-Based Flow
Data Grounding Decision
Complexity Assessment Flow
Budget & Timeline Tradeoffs
Governance & Compliance Path
Multi-Agent Orchestration
Upgrade Paths
Lifecycle Check
IQ Layer Selection

Diagram Index

Diagram	Purpose	Maps To Framework
1. Complete Decision Flow	End-to-end technology selection	Phase 2: Q1-Q9 - All nine critical questions
2. Data Grounding Decision	Data strategy and knowledge sources	Phase 2: Q3 - Data grounding patterns
3. Persona-Based Flow	Selection by role and skill level	Phase 2: Q2 - Build approach + Scenarios
4. Complexity Assessment	Technical complexity evaluation	Evaluation Criteria: Complexity
5. Budget & Timeline	Cost and time-to-production paths	Evaluation Criteria: Budget & Time
6. Governance & Compliance	Security and compliance requirements	Evaluation Criteria: Governance
7. Multi-Agent Orchestration	Multi-agent patterns and frameworks	Quick Reference: Orchestration Complexity
8. Upgrade Paths	Migration and progressive enhancement	Implementation Patterns: Progressive Enhancement
9. Lifecycle Check	Pre-flight readiness gate	Evaluation Criteria: Lifecycle & Operational Readiness
10. IQ Layer Selection	Knowledge grounding domain selection	Microsoft AI Stack: Microsoft IQ

Complete Decision Flow

Interactive version available. Pan, zoom, and click nodes in the embedded explorer below, or open the full explorer.

View as static Mermaid diagram

%%{init: {'theme':'dark'}}%%
flowchart TD
    Start([Start: Need AI Solution]) --> Q1{User interaction<br/>pattern?}

    Q1 -->|Conversational/Chat UI| UI_Path[UI-Based Agent]
    Q1 -->|Autonomous/Event-driven| Auto_Path[Autonomous Agent]
    Q1 -->|API/Headless Service| API_Path[Headless Service]

    UI_Path --> Q2{Where will users<br/>interact?}
    Q2 -->|M365 apps only| M365[M365 Copilot]
    Q2 -->|Teams + custom| Q3{Build approach?}
    Q2 -->|Multi-channel| Q3

    Q3 -->|Low-code| Studio[Copilot Studio]
    Q3 -->|Pro-code| Q4{Primary platform?}

    Q4 -->|M365-centric| SDK[M365 Agents SDK]
    Q4 -->|Azure-centric| Q4Azure{Need custom UI<br/>protocol?}

    Q4Azure -->|Yes| AGUI[Agent Framework + AG-UI<br/><i>Preview</i>]
    Q4Azure -->|No| Foundry[Microsoft Foundry - Azure]

    Auto_Path --> Q5{Build approach?}
    Q5 -->|Low-code| StudioAuto[Copilot Studio<br/>with Event Triggers]
    Q5 -->|Pro-code| Q6{Primary workflow type?}

    Q6 -->|Enterprise integration| LogicApps[Azure Logic Apps<br/>agentic workflows<br/><i>Consumption Preview</i>]
    Q6 -->|Custom orchestration| Foundry

    API_Path --> Q7{Hosting preference?}
    Q7 -->|Managed PaaS| AgentService[Foundry Agent Service]
    Q7 -->|Self-hosted| Foundry
    Q7 -->|Local/Edge| WinAI[Microsoft Foundry - Windows<br/><i>Local</i>]

    M365 --> DataQ{Need custom data?}
    Studio --> DataQ
    StudioAuto --> DataQ
    SDK --> DataQ
    AGUI --> DataQ
    Foundry --> DataQ
    LogicApps --> DataQ
    AgentService --> DataQ

    DataQ -->|M365 data| GraphConn[Graph Connectors]
    DataQ -->|Documents| AISearch[Azure AI Search<br/>Agentic retrieval GA via REST<br/>portal <i>Preview</i>]
    DataQ -->|Structured data| VectorDB{Vector DB choice?}
    DataQ -->|Analytics data| Fabric[Microsoft Fabric]
    DataQ -->|No grounding| DirectDeploy

    VectorDB -->|Global scale NoSQL| CosmosDB[Cosmos DB<br/>IVF/HNSW/DiskANN]
    VectorDB -->|Relational, OSS| PostgreSQL[PostgreSQL<br/>pgvector]
    VectorDB -->|SQL Database Engine| SQLFamily{Where does<br/>SQL run?}

    SQLFamily -->|Cloud PaaS| AzureSQL[Azure SQL Database<br/>VECTOR_SEARCH <i>Preview</i>]
    SQLFamily -->|Lift-and-shift PaaS| SQLMI[Azure SQL MI<br/>Verify vector features<br/>per update policy]
    SQLFamily -->|On-prem / VM| SQLServer[SQL Server 2025<br/>VECTOR_SEARCH <i>Preview</i><br/>PREVIEW_FEATURES required]
    SQLFamily -->|Fabric-native| SQLFabric[SQL database in Fabric<br/>VECTOR_SEARCH <i>Preview</i>]

    GraphConn --> DeployConfig
    AISearch --> DeployConfig
    CosmosDB --> DeployConfig
    PostgreSQL --> DeployConfig
    AzureSQL --> DeployConfig
    SQLMI --> DeployConfig
    SQLServer --> DeployConfig
    SQLFabric --> DeployConfig
    Fabric --> DeployConfig
    DirectDeploy --> DeployConfig

    DeployConfig[Configure Deployment] --> Q8{Deployment needs?}
    Q8 -->|M365 Copilot| Deploy_M365[Publish to M365 Copilot]
    Q8 -->|Teams| Deploy_Teams[Publish to Teams]
    Q8 -->|Web/Mobile| Deploy_Web[Web/Mobile Channels]
    Q8 -->|Multiple channels| Deploy_Multi[Multi-channel:<br/>10+ channels via SDK]
    Q8 -->|Azure service| Deploy_Azure[Azure Container Apps/<br/>App Service/AKS]

    Deploy_M365 --> Monitor[Monitor & Govern]
    Deploy_Teams --> Monitor
    Deploy_Web --> Monitor
    Deploy_Multi --> Monitor
    Deploy_Azure --> Monitor

    Monitor --> End([Production])

    style M365 fill:#004578,color:#fff
    style Studio fill:#004578,color:#fff
    style StudioAuto fill:#004578,color:#fff
    style SDK fill:#4b2070,color:#fff
    style Foundry fill:#a52617,color:#fff
    style AGUI fill:#4b2070,color:#fff
    style LogicApps fill:#004578,color:#fff
    style AgentService fill:#a52617,color:#fff
    style WinAI fill:#004578,color:#fff

Validation Summary

Last Validated: July 29, 2026

Read the status column, not the product name. Copilot Studio’s GitHub Copilot harness is GA as of 2026-08-03, as are generative orchestration (within the standard harness), computer use, A2A, and scheduled prompts. Its autonomous/triggered agents, child agents, connected agents, Foundry agents, and M365 Agents SDK agents are not documented as GA, and memory remains preview. Treat every trigger-driven design as Preview until you confirm the specific feature. (Copilot Studio what’s new)

UI-Based Agents (GA unless noted)

Technology	Action Safety	Proactive	Description
M365 Copilot	🔒 User-in-the-loop always	🔄 Reactive only	Conversational chat in M365 apps (docs)
Copilot Studio	⚠️ Actions can execute (add approval workflows)	🔄 Reactive (conversational) or ✅ Autonomous (event triggers)	Low-code, 13+ channels (docs)
M365 Agents SDK	⚠️ Custom action safety design	✅ Proactive capable	Pro-code, 10+ channels, C#/JS/Python, BYO orchestrator (docs)
Microsoft Foundry (Azure)	⚠️ Autonomous planning loops	✅ Proactive capable	Custom UI deployment (docs)
Agent Framework + AG-UI (Preview)	⚠️ Approvals via AG-UI middleware	✅ Proactive capable (inherits host orchestration)	Protocol bridges agents to web/mobile UI with SSE streaming, backend tool rendering, shared state, and CopilotKit components (docs)

Autonomous Agents

Technology	Action Safety	Proactive	Description
Copilot Studio (event triggers)	⚠️ Actions can execute	⚠️ Triggered execution, not documented as GA	Event triggers: SharePoint, OneDrive, Planner, Recurrence. Scheduled prompts reached GA 2026-07-01; triggered/autonomous agents carry no GA statement (docs)
Logic Apps agentic workflows	⚠️ Autonomous execution	✅ Proactive (event-driven)	Official terms are “agentic workflows” and “agent loop”. Consumption is explicitly in preview. Standard carries no preview banner on the agent loop, but Microsoft never states Standard is GA and specific Standard capabilities are marked preview. Check the exact capability (docs)
Microsoft Foundry (Azure) Agent Service	⚠️ Autonomous planning loops	✅ Proactive capable	Custom orchestration (docs)

API/Headless Services (GA)

Technology	Action Safety	Proactive	Description
Foundry Agent Service	⚠️ Autonomous planning loops	✅ Proactive capable	REST API, managed PaaS (docs)
Microsoft Foundry (Azure)	⚠️ Autonomous planning loops	✅ Proactive capable	REST API deployment (docs)

Vector Databases

Technology	Status	Capabilities
Cosmos DB	GA date not stated in docs	Flat/kNN, quantized flat, DiskANN index types; must be enabled as a feature. Ultra-high-throughput vector search is Private Preview (docs)
PostgreSQL pgvector	GA	`azure_ai` extension also enables in-database embedding generation and LLM calls (docs)
Azure SQL Database	`VECTOR_SEARCH()` Preview	Native `vector` type and `VECTOR_DISTANCE()` ship alongside it. Their status is documented separately, so verify before you assume. `TOP_N` is deprecated (docs)
Azure SQL MI	Not covered by the `VECTOR_SEARCH` preview note	Same SQL Database Engine; requires Always-up-to-date or SQL Server 2025 update policy. Verify vector feature availability for your update policy (docs)
SQL Server 2025	`VECTOR_SEARCH()` Preview	Additionally requires the `PREVIEW_FEATURES` database scoped configuration (docs)
SQL database in Fabric	`VECTOR_SEARCH()` Preview	Fabric-native SQL, same engine capabilities (docs)

Sources: Complete Decision Flow

Logic Apps agentic workflows concepts (verified 2026-07-29)
VECTOR_SEARCH T-SQL reference (Preview on Azure SQL DB, SQL database in Fabric, SQL Server 2025)
Azure AI Search agentic retrieval: GA via the 2026-04-01 REST API; portal experiences remain preview-only
Copilot Studio what’s new (verified 2026-07-29)

Persona-Based Flow

Interactive version available. Pan, zoom, and click nodes below, or open the full explorer.

View as static Mermaid diagram

%%{init: {'theme':'dark'}}%%
flowchart TD
    Start([Who are you?]) --> P1{Your role?}

    P1 -->|End user| User[Use M365 Copilot]
    P1 -->|Business maker| Maker{Dev support?}
    P1 -->|Developer| Dev{Focus area?}
    P1 -->|Data scientist/analyst| DS{Primary workload?}
    P1 -->|Integration specialist| IntSpec[Azure Logic Apps<br/>agentic workflows<br/><i>Consumption Preview</i>]

    Maker -->|No devs| MakerStudio[Copilot Studio]
    Maker -->|Occasional help| MakerPlus[Studio + Custom Actions]

    Dev -->|M365 integration| DevM365[M365 Agents SDK]
    Dev -->|Azure services| DevAzure[Microsoft Foundry - Azure]
    Dev -->|Multi-agent orchestration| DevOrch[Agent Framework]
    Dev -->|Multi-platform| DevChoice{Skill level?}
    Dev -->|Autonomous agents| DevAuto{Event-driven?}
    Dev -->|Custom UI streaming| DevAGUI[Agent Framework + AG-UI<br/><i>Preview</i>]
    Dev -->|My own codebase| DevLoop{Which rung of<br/>the developer loop?}

    DevLoop -->|Stay in the editor| DevIDE[GitHub Copilot agent mode<br/>in VS Code or Visual Studio]
    DevLoop -->|Delegate a whole issue| DevCloud[Copilot cloud agent<br/>GA on paid plans<br/>not on Copilot Free]
    DevLoop -->|Specialize the agent| DevCustom[Custom agents<br/>plus AGENTS.md<br/>agent instructions]
    DevLoop -->|Build on the harness| DevGHSDK[GitHub Copilot SDK<br/>GA - six languages<br/>Foundry hosting Python and C sharp only]

    DevChoice -->|Mid-level| DevStudio[Copilot Studio<br/>+ Custom Actions]
    DevChoice -->|Senior| DevSDK[M365 SDK or Foundry]

    DevAuto -->|Yes| DevLogic[Logic Apps<br/>agentic workflows<br/><i>Consumption Preview</i>]
    DevAuto -->|Custom orchestration| DevFoundry[Microsoft Foundry - Azure<br/>+ Agent Service]

    DS -->|Analytics/BI| DSFabric[Fabric data agents<br/>GA - needs F2+ or P1+]
    DS -->|ML/Custom models| DSFoundry[Microsoft Foundry - Azure]

    User --> UserEnd([Start using Copilot])
    MakerStudio --> MakerEnd([Build in Studio])
    MakerPlus --> MakerPlusEnd([Studio + code])
    DevM365 --> DevM365End([Build with SDK])
    DevAzure --> DevAzureEnd([Build in Foundry])
    DevOrch --> DevOrchEnd([Multi-agent workflows])
    DevStudio --> DevStudioEnd([Studio with code])
    DevSDK --> DevSDKEnd([Full code solution])
    DevLogic --> DevLogicEnd([Event-driven agent])
    DevFoundry --> DevFoundryEnd([Custom agent])
    DevAGUI --> DevAGUIEnd([Custom streaming UI])
    DevIDE --> DevIDEEnd([Inner loop assist])
    DevCloud --> DevCloudEnd([Issue in, pull request out])
    DevCustom --> DevCustomEnd([Repo-scoped specialists])
    DevGHSDK --> DevGHSDKEnd([Agent as a product])
    DSFabric --> DSFabricEnd([Data agent])
    DSFoundry --> DSFoundryEnd([ML pipeline])
    IntSpec --> IntSpecEnd([Enterprise workflows])

    style User fill:#0b6a0b,color:#fff
    style MakerStudio fill:#004578,color:#fff
    style MakerPlus fill:#004578,color:#fff
    style DevM365 fill:#4b2070,color:#fff
    style DevAzure fill:#4b2070,color:#fff
    style DevOrch fill:#4b2070,color:#fff
    style DevLogic fill:#004578,color:#fff
    style DevFoundry fill:#a52617,color:#fff
    style DevAGUI fill:#4b2070,color:#fff
    style DevIDE fill:#4b2070,color:#fff
    style DevCloud fill:#4b2070,color:#fff
    style DevCustom fill:#4b2070,color:#fff
    style DevGHSDK fill:#8c5e00,color:#fff
    style DSFabric fill:#8c5e00,color:#fff
    style DSFoundry fill:#a52617,color:#fff
    style IntSpec fill:#004578,color:#fff

Validation Summary - Persona-Based Flow

P1 is a node id, not a taxonomy. Microsoft does not use “1P/3P” as documented terminology. The documented ownership taxonomy is the Agent Registry four types: Microsoft agents (“built and maintained by Microsoft”) · External partner-built agents (“built by trusted non-Microsoft developers”) · Published by your org (“custom agents approved and published by your organization… might be referred to as LOB agents”) · Shared by creator (“created and shared by individual users or developers at your organization”). Learn those four; map the slang once if you must. (Agent Registry)

Last Validated: July 29, 2026

End User (GA)

Technology	Description	Documentation
M365 Copilot	Built-in AI in M365 apps, no setup required	M365 Copilot

Business Maker (GA)

Technology	Description	Documentation
Copilot Studio	Low-code platform, no dev support needed	Copilot Studio
Copilot Studio + Custom Actions	Low-code with occasional developer support for custom connectors/flows	Add tools to custom agents

Developer (GA unless noted)

Technology	Description	Documentation
M365 Agents SDK	Pro-code for M365-centric solutions, C#/JavaScript/Python, 10+ channels, BYO orchestrator	M365 Agents SDK
Microsoft Foundry (Azure)	Pro-code for Azure-centric solutions, custom models, full control	Microsoft Foundry
Microsoft Agent Framework	Microsoft’s investment direction - Microsoft describes it as the “direct successor” and “next generation of both” Semantic Kernel and AutoGen. Orchestration patterns: Sequential, Concurrent, Handoff, Group Chat, Magentic. Languages: Python, C#/.NET, and Go (Public Preview). No sunset date has been announced for Semantic Kernel or AutoGen.	Agent Framework
Agent Framework + AG-UI (Preview)	Protocol layer for web/mobile clients, supports SSE streaming, backend tool rendering, human approvals, shared/predictive state, and CopilotKit components.	AG-UI Integration
Copilot Studio + Custom Actions	Mid-level developers, low-code with custom code extensibility	Add tools to custom agents
Logic Apps agentic workflows	Event-driven agent loop, 1,400+ connectors. Consumption is explicitly in preview. Standard shows no preview banner on the agent loop, but Microsoft never states Standard is GA and some Standard capabilities are marked preview.	Logic Apps agentic workflows

Developer: The Codebase Loop (Bucket 3)

The Trade-off: Speed of the inner loop vs. ownership of the harness. The four rungs below are the developer-loop journey from the Capability Model. Each rung buys asynchrony and costs governance.

Rung	Technology	Status and limits	Documentation
1. Assist in the editor	GitHub Copilot agent mode	GA in VS Code and Visual Studio	Copilot agents in VS Code
2. Delegate the issue	Copilot cloud agent (renamed from “coding agent”)	GA on all paid plans including Student; NOT available on Copilot Free	About Copilot cloud agent
3. Specialize the agent	Custom agents (`.md` + YAML frontmatter) and `AGENTS.md`, which GitHub calls “agent instructions”	Custom agents are GA for the cloud agent, VS Code, and Visual Studio; Public Preview for JetBrains, Eclipse, and Xcode. GitHub’s own caveat on agent instructions: “currently not supported by all Copilot features.”	Custom agents
4. Build on the harness	GitHub Copilot SDK	GA. Python, TypeScript, Go, .NET, Java, Rust. Wraps the Copilot CLI engine over JSON-RPC. BYOK: OpenAI, Microsoft Foundry, Anthropic.	Copilot SDK

The seam has a hard edge. Microsoft Foundry Agent Service explicitly lists the GitHub Copilot SDK as a supported framework for Hosted agents, but Foundry Hosted agents support Python and C# only. The SDK ships six language bindings; Foundry hosting accepts two. A Go, Rust, Java, or TypeScript Copilot SDK agent is not directly hostable as a Foundry Hosted agent. Pick the language at rung 4 with rung 5 already in mind.

Business Builder: The Natural-Language Build Path (Bucket 3)

The Trade-off: Speed of the build vs. ownership of the app. The first three rows are alternative build paths, not steps. Cowork and Copilot Studio apps run on Copilot Managed Runtime, and Microsoft says Code’s shared apps use the same foundation; the Run it and Hand it over rows describe that host. The last row is Power Platform’s own path, which hands over by fork instead. The Copilot Studio and Power Apps vibe pages both warn: “Preview features aren’t meant for production use and might have restricted functionality.” How to choose among them is worked through in Scenario 11.

Step	Technology	Status and limits	Documentation
Describe it: just you, or a quick team tool	Copilot Code (Code in the Copilot app)	Frontier rollout; Microsoft states the timing two ways. Official Microsoft Blog, 2026-09-25: “Code is rolling out to Frontier at the end of the month, with broad availability in the coming weeks”; earlier in the same post, “Home and Code will start rolling out in our Frontier program in the coming weeks”. The same day’s runtime announcement: “Microsoft entry points are already live with Copilot Cowork, Copilot Code, and Copilot Studio”. Feature and rollout details are blog-sourced; no dedicated Learn feature overview was found as of 2026-09-25, though Learn documents Code’s usage-based billing on Copilot Credits. Learn’s list of runtime creation paths does not include Code.	Official Microsoft Blog \| Runtime announcement \| USL and UBB
Describe it: a team app over Microsoft 365 content	App skill in Copilot Cowork	Frontier program only. “Cowork requires a Microsoft Copilot license, but the subscription doesn’t include Cowork usage.” Shared by link. Cowork’s page: “Anyone with the link can open and use the app, including all its data”. The runtime FAQ: sharing “doesn’t grant access to the app’s underlying data”. The pages conflict; test with an account that lacks source access before sharing.	Use Copilot Cowork \| Admin overview
Build it: one piece of a business process	Copilot Studio apps	Preview. “Microsoft is gradually rolling out apps in Copilot Studio (Preview). This feature might not be available in your environment yet.” App creation is “powered by the GitHub Copilot harness”, and using, building, previewing, and testing apps can consume Copilot Credits. “The code view is read-only.” Sharing an app “might not automatically share its underlying data”. The apps documentation reviewed for this update gives no example of an app invoking an agent or agent flow.	Apps overview \| FAQ
Run it	Copilot Managed Runtime	Preview. A governed host, not a completed governance review. Microsoft Entra sign-in. Connector governance depends on the environment group: with an advanced connector policy and Advanced connector policies only off, “The most restrictive result applies”; with that setting on and no ACP, “Data policies are ignored. Without an ACP, connectors aren’t restricted.” Sharing limits per environment group. One Apps inventory in the Microsoft 365 admin center, whose Data & tools view “doesn’t provide a complete view of exact destinations, dynamically resolved endpoints, actions that the app executed, or each user’s permissions to the underlying data”. Apps must be TypeScript or JavaScript single-page apps. Running an app takes Copilot Credits or a Power Apps Premium license; with Premium, credits still apply to separately billed services such as Work IQ APIs, or past the plan’s API request limits. Git pushes and code changes are not audited in Purview. Automatic for eligible commercial-cloud tenants; sovereign clouds and GCC not stated as of 2026-09-25.	Overview \| Governance \| Security and compliance \| FAQ
Hand it over	GitHub Copilot with the Managed Runtime plugin, or the `ms` CLI	Preview (SDK, CLI, and plugin). Same Git repository, same runtime. The plugin “also works with Claude Code and other coding agents through Open Plugins”. Repository type is fixed when the app is created: platform-managed Git, or your own repository on GitHub.com or GitHub Enterprise Cloud (GitHub Enterprise Server and Azure DevOps aren’t supported). With your own repository, “your existing branch policies and pull request reviews continue to apply”.	SDK overview \| GitHub Copilot quickstart
Or describe it on the Power Platform side	Power Apps vibe	Preview. Builds “requirements, data, and full apps with generated code”. Not available in a default environment; US, Australia, Asia, and India regions only, English only. Its own page says of Cowork and Copilot Studio app building: “We recommend using these experiences moving forward.” Handoff is a fork: “Converting a vibe app creates a new code app. The code app is disconnected from the original plan.”	Overview \| Convert to a code app

The handoff has a blind spot. The Hand it over row continues the same repository; the Power Apps vibe row creates a separate code app. On the runtime, Purview lists Git pushes and code changes as “Not audited”. Bind the app to your own GitHub repository at creation and your branch policies and pull-request reviews keep applying. For the platform-managed default, the documentation we reviewed doesn’t establish a review gate, so define one on purpose before the handoff.

Data Scientist/Analyst

Technology	Description	Documentation
Fabric data agents	GA (formerly “AI skill”). Analytics/BI focus, Python SDK, evaluation capabilities, Power BI/semantic models. Requires F2+ or P1+ capacity.	Fabric data agent \| Python SDK
Microsoft Foundry (Azure)	ML/custom models, full AI/ML pipeline control	Microsoft Foundry

Integration Specialist

Technology	Description	Documentation
Logic Apps agentic workflows	Enterprise integration focus, 1,400+ connectors, workflow automation. Consumption agentic workflows are explicitly in preview; Standard has no preview banner on the agent loop, but Microsoft never states Standard is GA. Verify the exact capability you plan to use.	Logic Apps Overview \| Agentic workflows

Sources: Persona-Based Flow

About Copilot cloud agent (rename verified 2026-07-29)
Microsoft Foundry Agent Service: Hosted agents (Python and C# only)
Fabric data agent (GA; F2+ or P1+)
Logic Apps agentic workflows concepts (verified 2026-07-29)
What is Copilot Managed Runtime (preview) (prerelease documentation; three documented entry points: Cowork, Copilot Studio, SDK and CLI; verified 2026-09-25)
Copilot Managed Runtime default governance settings and security and compliance (lists “Git push / code change” as “Not audited”; verified 2026-09-25)
FAQ about Copilot Managed Runtime (sharing does not grant data access, which conflicts with the Cowork page; limits of the Data & tools view; runtime eligibility is Copilot Credits or Power Apps Premium; verified 2026-09-25)
Copilot Managed Runtime overview and key concepts for admins (Cowork build licensing, Power Apps Premium exclusions, creation-path defaults; verified 2026-09-25)
Copilot Managed Runtime SDK overview and GitHub Copilot quickstart (preview; repository options and the Claude Code note; verified 2026-09-25)
Use Copilot Cowork: build apps with the App skill (Frontier; page updated 2026-09-14, verified 2026-09-25)
Copilot Studio apps overview (preview, rolling out gradually; verified 2026-09-25) and FAQ for apps in Microsoft Copilot Studio (read-only code view; sharing might not share underlying data; verified 2026-09-25)
Overview of the new Power Apps vibe experience (preview; prerelease documentation updated 2026-09-08, with a banner recommending Cowork and Copilot Studio app building going forward; verified 2026-09-25) and Convert a vibe app to a code app (conversion creates a separate, disconnected code app; verified 2026-09-25)
Introducing the new Copilot with Home, Code and Autopilot (Official Microsoft Blog, Jared Spataro, 2026-09-25; Code’s feature description and rollout timing)
Microsoft Copilot Managed Runtime: Enterprise code execution for Microsoft 365 (Microsoft Copilot blog, 2026-09-25; entry points “already live” with Cowork, Code, and Copilot Studio)
Understanding the user subscription license (USL) and usage-based billing (UBB) (Code’s usage-based billing; verified 2026-09-25)

Data Grounding Decision

Interactive version available. Pan, zoom, and click nodes in the embedded explorer below, or open the full explorer.

View as static Mermaid diagram

%%{init: {'theme':'dark'}}%%
flowchart TD
    Start([Need to ground AI]) --> Q1{Where is<br/>your data?}

    Q1 -->|M365 only| M365Data{Data type?}
    Q1 -->|Azure only| AzureData{Data type?}
    Q1 -->|Both| Hybrid{Primary<br/>location?}
    Q1 -->|Analytics platform| FabricData[Microsoft Fabric]

    M365Data -->|SharePoint, OneDrive| Graph[Microsoft Graph Connectors]
    M365Data -->|Teams messages| GraphTeams[Graph Connectors]
    M365Data -->|Custom M365 app| GraphCustom[Custom Graph Connector]

    AzureData -->|Documents, PDFs| Q2{File count?}
    AzureData -->|Structured DB| DB{Database?}
    AzureData -->|Unstructured| Blob[Blob Storage with AI Search]

    Q2 -->|Under 10,000 files<br/>Simple setup| FileSearch{Platform?}
    Q2 -->|Production scale<br/>Custom chunking| Search[Azure AI Search]
    Q2 -->|Multimodal content| ContentUnderstanding[Azure AI Content Understanding<br/>with AI Search<br/>Preview]

    FileSearch -->|Azure Agent Service| AgentFileSearch[Agent Service File Search<br/>Auto chunk and embed]
    FileSearch -->|Copilot Studio| StudioKnowledge[Studio Knowledge Base<br/>Up to 1,000 files<br/>SharePoint or OneDrive]

    DB -->|Global scale, NoSQL| Cosmos{Vector algorithm?}
    DB -->|Relational, OSS| Postgres[PostgreSQL with pgvector]
    DB -->|SQL Database Engine| SQLFamily{Where does<br/>your SQL run?}

    SQLFamily -->|Cloud PaaS| AzureSQL[Azure SQL Database<br/>VECTOR_SEARCH Preview]
    SQLFamily -->|Lift-and-shift PaaS| SQLMI[Azure SQL MI<br/>Verify vector features<br/>per update policy]
    SQLFamily -->|On-prem / VM| SQL[SQL Server 2025<br/>VECTOR_SEARCH Preview<br/>PREVIEW_FEATURES required]
    SQLFamily -->|Fabric-native| SQLFabric[SQL database in Fabric<br/>VECTOR_SEARCH Preview]

    Cosmos -->|Flat index| CosmosIVF[Cosmos DB flat kNN]
    Cosmos -->|Quantized| CosmosHNSW[Cosmos DB quantized flat]
    Cosmos -->|Disk-optimized| CosmosDiskANN[Cosmos DB DiskANN]

    Hybrid -->|M365 primary| HybridM365[Graph with BYOK]
    Hybrid -->|Azure primary| HybridAzure[AI Search with Graph API]

    FabricData --> Q3{Access method?}
    Q3 -->|Direct platform access| FabricPlatform[Fabric Lakehouse and Warehouse<br/>OneLake with SQL endpoint]
    Q3 -->|Conversational agent| FabricAgent[Fabric data agents<br/>GA - needs F2+ or P1+]

    Graph --> Platform{Which<br/>platform?}
    GraphTeams --> Platform
    GraphCustom --> Platform
    AgentFileSearch --> Platform
    StudioKnowledge --> Platform
    Search --> Platform
    ContentUnderstanding --> Platform
    CosmosIVF --> Platform
    CosmosHNSW --> Platform
    CosmosDiskANN --> Platform
    Postgres --> Platform
    AzureSQL --> Platform
    SQLMI --> Platform
    SQL --> Platform
    SQLFabric --> Platform
    Blob --> Platform
    HybridM365 --> Platform
    HybridAzure --> Platform
    FabricPlatform --> Platform
    FabricAgent --> Platform

    Platform -->|M365 Copilot| PlatformM365([Use built-in])
    Platform -->|Copilot Studio| PlatformStudio([Configure Studio])
    Platform -->|M365 SDK| PlatformSDK([Code integration])
    Platform -->|Microsoft Foundry Azure| PlatformFoundry([Configure Foundry])
    Platform -->|Logic Apps| PlatformLogicApps([MCP Server<br/>Connector])

    style Graph fill:#004578,color:#fff
    style AgentFileSearch fill:#004578,color:#fff
    style StudioKnowledge fill:#4b2070,color:#fff
    style Search fill:#004578,color:#fff
    style ContentUnderstanding fill:#006192,color:#fff
    style CosmosIVF fill:#004578,color:#fff
    style CosmosHNSW fill:#004578,color:#fff
    style CosmosDiskANN fill:#004578,color:#fff
    style Postgres fill:#004578,color:#fff
    style AzureSQL fill:#004578,color:#fff
    style SQLMI fill:#004578,color:#fff
    style SQL fill:#004578,color:#fff
    style SQLFabric fill:#8c5e00,color:#fff
    style FabricPlatform fill:#8c5e00,color:#fff
    style FabricAgent fill:#8c5e00,color:#fff

Validation Summary - Data Grounding Decision

Last Validated: July 29, 2026

M365 Data Sources (GA)

Technology	Capabilities	Documentation
Microsoft Graph Connectors	M365 data sources (SharePoint, OneDrive, Teams)	Graph Connectors Overview

Document Processing - File Search (GA)

Technology	Capabilities	Documentation
Foundry Agent Service File Search Tool	Built-in file search with automatic parsing, chunking (800 tokens/400 overlap), embedding (text-embedding-3-large), keyword + semantic search, reranking. Supports up to 10,000 files per vector store (max 512 MB/file). Two modes: Basic (Microsoft-managed) vs Standard (BYO Azure AI Search + Blob Storage). Supported formats: .doc, .docx, .pdf, .pptx, .py, .md, .txt, .json, .html, .java, .cs, .cpp, and more. Service handles entire ingestion automatically.	Agent Service File Search
Copilot Studio Knowledge Base	File upload from local/OneDrive/SharePoint. Supports .doc, .docx, .ppt, .pptx, .pdf, .xls, .xlsx, .txt, .md, .html, .csv, .xml. Max 512 MB per file. Direct uploads allow up to 500 files per agent, while each SharePoint/OneDrive document source supports up to 1,000 files. Automatic chunking and vectorization into Dataverse with semantic indexing. OneDrive/SharePoint: Auto-sync (updates reflected automatically) vs Upload: Static files. SharePoint: User-scoped permissions (only files user has access to).	Copilot Studio Knowledge \| File Upload \| SharePoint Files \| Quotas and limits

Document Processing - Production Scale (GA)

Technology	Capabilities	Documentation
Azure AI Search	Document indexing, full-text search, vector search, hybrid queries, custom chunking strategies (fixed-size, variable-size, Document Layout skill). Requires manual setup of indexers, skillsets, chunking strategy. Production-scale scenarios with millions of documents.	AI Search Overview \| Chunking Strategies

Document Processing - Multimodal (Preview)

Technology	Capabilities	Documentation
Azure AI Content Understanding	Multimodal processing (documents/images/audio/video), RAG-ready Markdown output, AI Search custom skill integration, built-in chunking, standard/pro modes. API version: 2025-05-01-preview.	Content Understanding Overview \| Multimodal Search

Structured Databases - Vector Search

Technology	Status	Capabilities	Documentation
Cosmos DB Vector Search	GA date is not stated in the docs. Treat the silence as silence	Three index types: flat/kNN, quantized flat, DiskANN. Must be enabled as a feature. Ultra-high-throughput vector search is Private Preview	Cosmos DB Vector Search
PostgreSQL pgvector	GA	The `azure_ai` extension additionally enables in-database embedding generation and LLM calls. There is no “PostgreSQL agent” product	PostgreSQL Vector Search
Azure SQL Database	`VECTOR_SEARCH()` Preview	Native `vector` type and `VECTOR_DISTANCE()` are documented separately. Do not assume they inherit the same status. `TOP_N` is deprecated	VECTOR_SEARCH
Azure SQL Managed Instance	Not named in the `VECTOR_SEARCH` preview note	Same SQL Database Engine; requires Always-up-to-date or SQL Server 2025 update policy. Verify feature availability for your policy	SQL MI Update Policy
SQL Server 2025	`VECTOR_SEARCH()` Preview	Additionally requires the `PREVIEW_FEATURES` database scoped configuration	SQL Server Vector
SQL database in Fabric	`VECTOR_SEARCH()` Preview	Fabric-native SQL, same engine as Azure SQL Database	SQL database in Fabric

Analytics Platform (GA with Preview features)

Technology	Status	Capabilities	Documentation
Microsoft Fabric Platform	GA	Direct knowledge source access via Lakehouse (Delta tables, Spark), Warehouse (T-SQL), OneLake (ADLS Gen2 APIs), KQL databases. Microsoft Foundry (Azure) integration for RAG	Fabric Overview \| Microsoft Foundry (Azure) FAQ
Fabric data agents	GA	Formerly “AI skill”. Analytics data grounding (warehouses, lakehouses, Power BI semantic models, KQL databases), consumable from Copilot Studio and Foundry Agent Service. Requires F2+ or P1+ capacity	Fabric data agent \| Copilot Studio Integration
Fabric IQ	Preview	Documented Fabric workload for ontologies, graph models, semantic models, and operations agents, a different job from the data agent	Fabric IQ
Foundry Tools in Fabric	Preview	Billed to Fabric capacity units	Fabric Overview

Sources: Data Grounding Decision

Azure AI Search agentic retrieval: GA via the 2026-04-01 REST API; portal experiences remain preview-only. The 2026-05-01-preview adds remote sources: MCP Server, Work IQ, Fabric data agent, Fabric ontology. Azure AI Search was not renamed in the Foundry rebrand
VECTOR_SEARCH T-SQL reference (Preview)
Fabric data agent (GA) and Fabric IQ (Preview), verified 2026-07-29

MCP Integration (Preview)

Technology	Capabilities	Documentation
Logic Apps MCP Server	Standard logic apps as remote MCP servers, 1,400+ connectors, OAuth 2.0 auth, Streamable HTTP/SSE transports	Logic Apps MCP Server \| API Center Integration

When to Use File Search vs Azure AI Search

Solution	Use When	Cost
File Search (Agent Service/Copilot Studio)	Up to 10,000 files, simple setup (no manual indexer/chunking config), automatic embedding, suitable for smaller document sets, internal knowledge bases, rapid prototyping	Included in Agent Service consumption or Copilot Studio credits
Azure AI Search	Production scale (millions of documents), custom chunking strategies required, advanced features (analyzers, scoring profiles, faceting), complex indexing pipelines, enterprise search	Dedicated AI Search tier (Basic ~$75/mo to S3 ~$3K/mo)

Complexity Assessment Flow

Interactive version available. Pan, zoom, and click nodes below, or open the full explorer.

View as static Mermaid diagram

%%{init: {'theme':'dark'}}%%
flowchart TD
    Start([Assess Use Case]) --> Q1{How many<br/>data sources?}

    Q1 -->|1| Simple
    Q1 -->|2 to 3| Q2{Workflow<br/>complexity?}
    Q1 -->|4 or more| Complex

    Q2 -->|Linear| Medium
    Q2 -->|Branching| Complex

    Simple[Low Complexity] --> S_Tech{Skills?}
    S_Tech -->|Makers| S_M365[M365 Copilot with Graph]
    S_Tech -->|Devs| S_Studio[Copilot Studio]

    Medium[Medium Complexity] --> M_Tech{Skills?}
    M_Tech -->|Makers| M_Studio[Copilot Studio]
    M_Tech -->|Devs| M_SDK[Studio with Custom Actions]

    Complex[High Complexity] --> C_Tech{Need custom<br/>models?}
    C_Tech -->|No| C_SDK[M365 Agents SDK]
    C_Tech -->|Yes| C_Foundry[Microsoft Foundry Azure]

    S_M365 --> Timeline1[Days to deploy]
    S_Studio --> Timeline2[1 to 2 weeks]
    M_Studio --> Timeline3[2 to 4 weeks]
    M_SDK --> Timeline4[1 to 2 months]
    C_SDK --> Timeline5[2 to 3 months]
    C_Foundry --> Timeline6[3 to 6 months]

    style Simple fill:#0b6a0b,color:#fff
    style Medium fill:#8c5e00,color:#fff
    style Complex fill:#a52617,color:#fff

Budget & Timeline Tradeoffs

Interactive version available. Pan, zoom, and click nodes below, or open the full explorer.

View as static Mermaid diagram

%%{init: {'theme':'dark'}}%%
flowchart TD
    Start([Project Constraints]) --> Q1{AI setup<br/>budget?}

    Q1 -->|Existing M365| M365Only[M365-Only Path]
    Q1 -->|USD 200 to 500 per month| Starter[Starter Budget]
    Q1 -->|USD 1K to 5K per month| Growth[Growth Budget]
    Q1 -->|USD 5K or more per month| Enterprise[Enterprise Budget]

    M365Only --> M365T{Timeline?}
    M365T -->|Days| M365_Fast[M365 Copilot Chat<br/>0 addl cost included with M365<br/>Instruction-based agents]
    M365T -->|1 to 2 Weeks| M365_Med[M365 Copilot with Graph<br/>0 addl cost<br/>Knowledge grounding]
M365T -->|1 Month or more| M365_Slow[Declarative Agents<br/>0 addl cost or PAYG<br/>Custom instructions and data]

    Starter --> StarterT{Timeline?}
    StarterT -->|1 to 2 Weeks| S_Fast[Copilot Studio PAYG<br/>USD 200 to 500 per month avg<br/>USD 0.01 per credit, low-code]
    StarterT -->|1 to 2 Months| S_Med[Studio with AI Builder<br/>USD 200 to 500 per month<br/>Document processing]
    StarterT -->|3 or more Months| S_Slow[Logic Apps agentic workflows<br/>Approx 200 to 400 per month<br/>Consumption Preview]

    Growth --> GrowthT{Timeline?}
    GrowthT -->|2 to 4 Weeks| G_Fast[Copilot Studio Capacity<br/>200 or more per month prepaid<br/>25K credits per pack]
    GrowthT -->|1 to 3 Months| G_Med[M365 SDK with Azure Basic<br/>USD 1 to 3K per month<br/>Custom agents, Azure hosting]
    GrowthT -->|3 to 6 Months| G_Slow[Microsoft Foundry Azure Starter<br/>USD 1 to 5K per month<br/>PAYG tokens with AI Search Basic]

    Enterprise --> EntT{Timeline?}
    EntT -->|4 to 8 Weeks| E_Fast[Foundry Serverless<br/>Approx 5 to 15K per month est<br/>PAYG with AI Search Standard]
    EntT -->|3 to 6 Months| E_Med[Foundry with Agent Service<br/>USD 10 to 30K per month est<br/>Managed orchestration]
    EntT -->|6 or more Months| E_Slow[Foundry PTU with Premium<br/>30K or more per month<br/>PTU reservation with S2 or S3 Search]

    style M365_Fast fill:#0b6a0b,color:#fff
    style M365_Med fill:#0b6a0b,color:#fff
    style S_Fast fill:#0b6a0b,color:#fff
    style G_Fast fill:#0b6a0b,color:#fff
    style G_Med fill:#0b6a0b,color:#fff
    style E_Med fill:#0b6a0b,color:#fff

    style M365_Slow fill:#8c5e00,color:#fff
    style S_Med fill:#8c5e00,color:#fff
    style G_Slow fill:#8c5e00,color:#fff
    style S_Slow fill:#9a4300,color:#fff

Validation Summary - Budget & Timeline Tradeoffs

Last Validated: July 29, 2026

M365-Only ($0 AI infrastructure add’l)

Solution	Monthly Cost	Documentation
M365 Copilot Chat	Free (included) with eligible M365 subscription	Web-grounded chat and instruction-based agents (docs)
M365 Copilot + Graph Connectors	$30/user/month M365 Copilot license	Graph Connectors included at no extra charge (docs)
Declarative Agents	$0 or PAYG	Instruction-based or public-web grounded = $0; shared tenant data = PAYG (docs)

Starter ($200-500/mo)

Solution	Monthly Cost	Documentation
Copilot Studio PAYG	$200-500/mo (typical)	$0.01 per Copilot Credit (docs)
AI Builder	Included in Power Platform	Document processing (invoices, receipts, contracts) (docs)
Logic Apps agentic workflows	$200-400/mo (typical)	Consumption agentic workflows are explicitly in preview. Event-driven agent loop (docs)

Growth ($1K-5K/mo)

Solution	Monthly Cost	Documentation
Copilot Studio Capacity Packs	$200/month per 25,000 credits	Prepaid (docs)
M365 SDK + Azure	$600-2.3K/mo	SDK free; Azure hosting (App Service ~$100-300/mo) + Azure OpenAI PAYG (~$500-2K/mo tokens) (docs)
Microsoft Foundry (Azure) Starter	$1-5K/mo estimate	PAYG tokens + AI Search Basic (~$75/mo) (OpenAI pricing) \| (AI Search pricing)

Enterprise ($5K+/mo)

Solution	Monthly Cost	Documentation
Foundry Serverless	$5-15K/mo	PAYG tokens at scale + AI Search Standard S1 (~$250/mo) (docs)
Foundry + Agent Service	$10-30K/mo	Managed orchestration PaaS + AI Search S2 (~$1K/mo) (docs)
Foundry PTU + Premium	$30K+/mo	PTU reservations (50+ PTUs minimum) + AI Search S2/S3 (docs)

Timeline Estimates

Timeline	Use Case	Example Scenario
Days	M365 built-in features, no development	(scenarios)
1-2 Weeks	Low-code platforms (Copilot Studio Employee Self-Service, Logic Apps)	(HR Knowledge Base scenario)
1-3 Months	Custom agents with SDKs, moderate complexity	(Customer Support scenario)
3-6 Months	Microsoft Foundry (Azure) custom solutions, complex orchestration	(evaluation-criteria)
6+ Months	Enterprise-scale with PTU, fine-tuning, advanced patterns

Cross-Tier: The Two Prepurchase Plans

The Trade-off: One invoice vs. two different currencies. There is no single “unified” prepurchase. Microsoft ships two distinct Azure Reservations, and buying the wrong one is an expensive way to learn the difference.

One plan covers Copilot Credit usage only; the other spans Foundry, Copilot Studio, Fabric, and GitHub. Both are prepaid, both are non-refundable, and the narrow one always burns first. See Microsoft AI Stack for why that ordering decides what a broad commitment is actually worth.

	Copilot Credit Pre-Purchase Plan	Microsoft Agent Prepurchase Plan
Unit	CCCU (Copilot Credit CUs)	ACU (Agent CUs)
Covers	“eligible Copilot Credit usage”	“select services across Microsoft Foundry, Microsoft Copilot Studio, Microsoft Fabric, and GitHub costs”, where the Copilot Studio entry also spans Dynamics 365 first-party agents and Copilot
Value	1 CCCU pays down US$1 of qualifying retail cost, purchased at a tiered discount (~5 to 6% in Microsoft’s worked examples)	1 ACU pays down US$1 of qualifying retail cost, purchased at a tiered discount (~5 to 6% in Microsoft’s worked examples)
Docs	Copilot Credit Pre-Purchase Plan	Microsoft Agent Prepurchase Plan

Both are 1-year terms with auto-renew on by default, and all purchases are final: no cancel, exchange, split, or merge.

Precedence: narrow benefits burn before broad benefits. Microsoft states it plainly: “Reservations always apply before prepurchase plans.” The order is:

Microsoft Foundry PTU Reservations
Microsoft Fabric Capacity Reservations
Copilot Credit Prepurchase Plan
Microsoft Agent Prepurchase Plan: “Applied last to remaining AI usage across all platforms.”

Do not assume coverage. Whether Copilot Cowork and Work IQ consumption draws down these plans is inferred, not stated in Microsoft’s documentation. Confirm with your account team before you model it into a business case.

See Evaluation Criteria: Budget.

Cost Calculation Notes

M365 per-user costs ($30/user/month) NOT included in bands - these are AI infrastructure costs only
Estimates assume moderate usage (not high-scale production)
Azure consumption highly variable based on tokens, requests, storage
PTU (Provisioned Throughput Units) require Azure Reservations for cost optimization

Sources

Copilot Studio Licensing (Updated: 2025-11-05)
M365 Copilot Cost Considerations (Updated: 2025-11-25)
Azure OpenAI Pricing (Updated: 2026)
Azure AI Search Tiers (Updated: 2025-11-06)
Logic Apps agentic workflows (Consumption Preview, verified 2026-07-29)
Microsoft Agent Prepurchase Plan and Copilot Credit Pre-Purchase Plan (both ms.date 2026-07-17)
AI Builder Overview (Updated: 2026-01-14)

Governance & Compliance Path

Interactive version available. Pan, zoom, and click nodes below, or open the full explorer.

View as static Mermaid diagram

%%{init: {'theme':'dark'}}%%
flowchart TD
    Start([Governance Requirements]) --> Q1{Data residency<br/>requirement?}

    Q1 -->|M365 tenant only<br/>Strict boundary| R1[M365 Tenant Boundary]
    Q1 -->|Azure region<br/>Configurable| R2[Azure Region Control]
    Q1 -->|Flexible<br/>Multi-region OK| R3[Flexible Residency]

    R1 --> Q2{Built-in DLP<br/>sufficient?}
    Q2 -->|Yes, inherit M365| G1[M365 Copilot<br/>M365 Boundary with DLP]
    Q2 -->|Need custom agents| G2[Copilot Studio<br/>Power Platform Governance]

    R2 --> Q3{Network<br/>isolation?}
    Q3 -->|VNet with private<br/>endpoints required| G3[Microsoft Foundry Azure<br/>VNet with Private Endpoints]
    Q3 -->|Managed runtime<br/>with VNet needed| G4[AI Agent Service<br/>Full Azure Governance]
    Q3 -->|Power Platform<br/>governance sufficient| G5[AI Builder<br/>Power Platform DLP]

    R3 --> Q4{Development<br/>model?}
    Q4 -->|Enterprise workflows<br/>with integration| G6[Logic Apps Standard<br/>Azure RBAC with VNet]
    Q4 -->|Multi-channel<br/>custom agents| G7[M365 Agents SDK<br/>Custom Auth with Hosting]
    Q4 -->|Orchestration<br/>library only| G8[Agent Framework<br/>Host inherited governance]

    style G1 fill:#0b6a0b,color:#fff
    style G2 fill:#004578,color:#fff
    style G3 fill:#004578,color:#fff
    style G4 fill:#004578,color:#fff
    style G5 fill:#8c5e00,color:#fff
    style G6 fill:#004578,color:#fff
    style G7 fill:#4b2070,color:#fff
    style G8 fill:#a52617,color:#fff

Validation Summary: Governance & Compliance Path

Last Validated: July 29, 2026

M365 Tenant Boundary Technologies

Technology	Action Safety	Proactive	Data Grounding/Memory/Analytics	Key Governance Features	Documentation
M365 Copilot (GA)	✅ User-in-the-loop (drafts only)	❌ Reactive only	Grounding only (M365 Graph per-request), no extractable memory	M365 trust boundary, auto DLP/sensitivity labels, user-scoped permissions, Purview audit, no training on tenant data	M365 Copilot Security
Copilot Studio (GA)	⚠️ Actions execute (flows/connectors)	✅ Autonomous (event triggers)	Grounding + Dataverse persistence (transcripts, variables), full analytics	Power Platform RBAC + DLP, environment-level governance, connector controls, ⚠️ web search leaves boundary, ⚠️ dual auth (user/service)	Copilot Studio Security

Azure Region Control Technologies

Technology	Action Safety	Proactive	Data Grounding/Memory/Analytics	Key Governance Features	Documentation
Microsoft Foundry (Azure) (GA)	⚠️ Tool calling with autonomous planning	✅ Proactive (Azure Functions, Logic Apps)	Grounding + BYO thread storage (Cosmos DB), Azure Monitor + OpenTelemetry	Azure RBAC (control + data plane), VNet + private endpoints, managed identity, CMK optional, ⚠️ API key OR Entra ID (Entra recommended)	Microsoft Foundry security
AI Agent Service	⚠️ Autonomous with action tools (Logic Apps, Functions, MCP)	✅ Proactive (event triggers)	BYO thread storage (Cosmos DB), Azure Monitor project-scoped	Full RBAC (project + resource), VNet + private endpoints, BYO storage, no public egress by default, container injection for VNet	Agent Service Security
AI Builder (GA)	Varies by model type	❌ Reactive	Dataverse storage	Power Platform DLP, Dataverse RBAC, environment location	AI Builder Governance

Flexible Residency Technologies

Technology	Action Safety	Proactive	Data Grounding/Memory/Analytics	Key Governance Features	Documentation
Logic Apps Standard (workflow platform)	⚠️ Autonomous execution (workflows)	✅ Proactive (event-driven, 1,400+ connectors)	Workflow state + connector data	Granular Azure RBAC, VNet + private endpoints (Standard), managed identity, Customer Lockbox, FedRAMP/HIPAA/ISO 27001	Logic Apps Security
M365 Agents SDK (GA)	⚠️ Custom design (full developer responsibility)	✅ Proactive (custom event handling)	Custom (developer implements)	Custom auth (MSAL, Entra ID), hosting platform RBAC, ⚠️ delegated OR application permissions, self-hosted = full network control	M365 SDK Auth
Agent Framework (GA)	Inherits from host app	Inherits from host app	Inherits from host app	No built-in governance (library only), inherits from hosting platform	Agent Framework

Key Decision Criteria:

M365 tenant boundary required → M365 Copilot (strict) or Copilot Studio (custom agents)
Azure region control + VNet isolation → Microsoft Foundry (Azure) or Foundry Agent Service
Enterprise workflow automation → Logic Apps Standard
Custom multi-channel agents → M365 Agents SDK (full control, full responsibility)
Orchestration library only → Agent Framework (no built-in governance)

The Logic Apps status trap. The rows above describe the workflow platform, not the agent loop. Microsoft explicitly marks Consumption agentic workflows as preview. Standard carries no preview banner on the agent loop itself, but Microsoft never states Standard is GA, and specific Standard capabilities are marked preview. Absence of a banner is silence, not a guarantee. Check the exact capability you plan to ship. (Agentic workflows concepts, verified 2026-07-29)

Identity for agents has a product name now. Microsoft Entra Agent ID is a formally named product within Microsoft Entra that provides the platform for creating and managing agent identities and agent identity blueprints, and Microsoft states it is now generally available. Microsoft Agent 365 reached GA on 2026-05-01, is licensed per user, and is included in Microsoft 365 E7; its Shadow AI component remains preview. (Entra Agent ID | Agent 365 overview)

Multi-Agent Orchestration

Interactive version available. Pan, zoom, and click nodes below, or open the full explorer.

View as static Mermaid diagram

%%{init: {'theme':'dark'}}%%
flowchart TD
    Start([Need coordination?]) --> IsMulti{More than one agent?}
    IsMulti -->|No| SingleExit[Single-agent event trigger<br/>Not multi-agent orchestration]
    IsMulti -->|Yes| Q1{Pattern Type?}

    Q1 -->|Connected agents<br/>Mesh A2A| Connected[Connected mesh pattern]
    Q1 -->|Sequential or parallel<br/>workflows| Workflows[Agent Workflow Orchestration]

    Connected --> C_Platform{Platform?}
    C_Platform -->|Low-code| C_Studio[Copilot Studio<br/>A2A GA April 2026<br/>Connect other agents <i>Preview</i>]
    C_Platform -->|Azure| C_Foundry[Foundry Agent Service<br/>Incoming A2A endpoint<br/><i>Preview</i>]

    Workflows --> W_Framework{Framework?}
    W_Framework -->|Microsoft| W_AgentFW[Microsoft Agent Framework<br/>Python and .NET GA<br/>Go <i>Public Preview</i>]
    W_Framework -->|Open source| W_SK[Semantic Kernel<br/>Superseded by Agent Framework<br/>no sunset date announced]
    W_Framework -->|Bring your own| W_SDK[M365 Agents SDK<br/>Integrate Agent Framework or SK]
    W_Framework -->|Third-party state| W_LangGraph[LangGraph Third-Party<br/>State graphs]

    C_Studio --> Note1[Can connect Fabric data agents<br/>GA as data grounding participants]
    C_Foundry --> Note2[Direct delegation is not<br/>deterministic workflow orchestration]

    C_Studio --> Deploy1([Deploy])
    C_Foundry --> Deploy2([Deploy])
    W_AgentFW --> Deploy3([Deploy])
    W_SK --> Deploy4([Deploy])
    W_SDK --> Deploy5([Deploy])
    W_LangGraph --> Deploy6([Deploy])
    SingleExit --> Deploy7([Choose single-agent trigger path])

    style Connected fill:#0b6a0b,color:#fff
    style Workflows fill:#8c5e00,color:#fff
    style SingleExit fill:#4b2070,color:#fff

Validation Summary: Multi-Agent Orchestration

Last Validated: July 29, 2026

Connected Agents / Sub-Agent Pattern

Technology	Status	Capabilities	Documentation
Copilot Studio	A2A GA (April 2026); “Connect other agents” is Preview	Agent-to-agent mesh, connected agents, and what Microsoft’s guidance calls “inline agents, also known as child agents.” Do not read A2A’s GA as a blanket GA for the multi-agent surface	Connected Agents
Foundry Agent Service	Preview (incoming A2A endpoint)	Lightweight direct delegation; direction and protocol support must be validated	A2A integration
Fabric data agents	GA	Consumed by other agents for data grounding (NOT an orchestrator). Requires F2+ or P1+ capacity	Fabric integration

Agent Workflow Orchestration

Foundry Workflows is retiring. Microsoft’s wording, quoted exactly:

“Microsoft Foundry is retiring workflows on December 1, 2026. If you’re looking to build new workflows, use Microsoft Agent Framework. To migrate existing workflows, see the Migration guide section of this article for all supported paths.”

Three documented migration paths, each with a different center of gravity:

Microsoft Agent Framework: the recommended path; exported YAML is reusable with minimal changes.
Azure Logic Apps: keeps a visual designer, aimed at business-process automation.
Agent-to-agent (A2A): lightweight handoffs.

Exported YAML definitions remain executable when deployed as Hosted Agents. After 2026-12-01, the visual designer and in-portal execution are unsupported.

Technology	Status	Orchestration Patterns	Documentation
Microsoft Agent Framework	GA (Python, C#/.NET); Go is Public Preview	Microsoft’s stated “direct successor” to Semantic Kernel and AutoGen: Sequential, Concurrent, Handoff, Group Chat, Magentic	Agent Framework
Semantic Kernel	Superseded by Agent Framework; no sunset date announced	Sequential, Concurrent, Group Chat, Handoff, Magentic. Migration guides exist for both SK and AutoGen	Semantic Kernel Agents
M365 Agents SDK	GA	BYO orchestrator (integrate Agent Framework or third-party)	M365 SDK
LangGraph	Third-party	State graph management for complex workflows	Third-party framework

Event-Driven Agents (Single Agent Triggering)

Technology	Status	Event Handling	Documentation
Logic Apps agentic workflows	Consumption explicitly Preview; Standard has no preview banner on the agent loop, but Microsoft never states Standard is GA	Event triggers + MCP Server (triggers a SINGLE agent, NOT multi-agent orchestration)	Agentic workflows
Azure Functions + Agent Service	GA	Event-driven single agent invocation (event routing, NOT coordination)	Agent Service
Event Grid + Foundry	GA	Event routing to trigger agents independently (NOT orchestration)	Azure Event Grid

Sources: Multi-Agent Orchestration

Foundry Workflows retirement and migration guide (retirement date 2026-12-01)
Microsoft Agent Framework (Python, C#/.NET; Go Public Preview)
Copilot Studio what’s new: A2A GA April 2026; “Connect other agents (Preview)” still listed (verified 2026-07-29)

Key Distinctions:

Connected/Mesh: Agents discover and invoke each other (A2A) or delegate to inline/child agents
Workflow Orchestration: Sequential/parallel/concurrent agent coordination patterns
Event-Driven: Single agent triggered by events (NOT multi-agent orchestration)

Upgrade Paths

Interactive version available. Pan, zoom, and click nodes below, or open the full explorer.

View as static Mermaid diagram

%%{init: {'theme':'dark'}}%%
flowchart TD
    Start([Start: Low-Code Agent]) --> Q1{Need more<br/>capability?}

    Q1 -->|No| Stay[Copilot Studio<br/>Standard]
    Q1 -->|Yes| Q2{What kind of<br/>capability?}

    Q2 -->|Better Reasoning/LLM| Path1[Enhance<br/>BYO Model - Foundry<br/><i>Preview</i>]
    Q2 -->|Complex Orchestration| Path2[Extend<br/>Handoff to Agent Service<br/><i>Preview</i>]
    Q2 -->|Custom UI/Protocol| Path3[Migrate<br/>M365 Agents SDK<br/>or Foundry]

    Path1 --> Result1([Copilot Studio<br/>+ Specialized Model])
    Path2 --> Result2([Copilot Studio UI<br/>+ Azure Agent Brain])
    Path3 --> Result3([Full Pro-Code<br/>Solution])

    style Stay fill:#004578,color:#fff
    style Path1 fill:#8c5e00,color:#fff
    style Path2 fill:#a52617,color:#fff
    style Path3 fill:#4b2070,color:#fff

Validation Summary: Upgrade Paths

Last Validated: July 29, 2026

Progressive Enhancement Mechanisms

Mechanism	Status	Description	Documentation
BYO Model	Preview	Swap the default Copilot Studio model for a specialized model from the Microsoft Foundry catalog, which lists over 10,000 models	BYO Model
Agent Handoff	Preview	Copilot Studio delegates to a Foundry agent for complex tasks. Foundry agents are not documented as GA in Copilot Studio	Connect to a Foundry agent
Shared Knowledge	GA	Both Copilot Studio and custom apps can consume the same Azure AI Search index. Azure AI Search was not renamed in the Foundry rebrand	Azure AI Search

Using This Guide

How to navigate: Start with Complete Decision Flow for end-to-end guidance, or jump to specific sections: Persona-Based Flow (by role), Data Grounding (data architecture), Complexity Assessment (effort estimate), Budget & Timeline (cost/time), Governance Path (compliance), or Multi-Agent (orchestration).

Diagram conventions: Diamonds = decisions, Rectangles = technologies, Circles = start/end. Colors: Blue = Microsoft core, Purple = developer tools, Green = low complexity, Orange = medium complexity, Red = high complexity/enterprise.

Lifecycle Check

Interactive version available. Pan, zoom, and click nodes below, or open the full explorer.

View as static Mermaid diagram

%%{init: {'theme':'dark'}}%%
flowchart TD
    Start([Start: Pre-Flight Check]) --> Q0{Technology status?}

    Q0 -->|GA| GA_Path[✅ Production-safe]
    Q0 -->|Preview / RC| Preview_Path[⚠️ Plan for change]
    Q0 -->|Deprecated| Dep_Path[🚫 Identify successor]

    GA_Path --> Timeline{Retirement date<br/>within horizon?}
    Preview_Path --> Timeline
    Dep_Path --> Successor{Successor<br/>identified?}

    Timeline -->|No deadline| Proceed[✅ Proceed to<br/>Decision Framework]
    Timeline -->|Deadline exists| Urgency{Go-live BEFORE<br/>deadline?}

    Urgency -->|Yes| Proceed
    Urgency -->|No| Migrate[⚠️ Build on successor<br/>or migrate first]

    Successor -->|Yes| Migrate
    Successor -->|No| Block[🚫 Block. Research<br/>successor path]

    Migrate --> Proceed

    style GA_Path fill:#1a6b3c,color:#fff
    style Proceed fill:#1a6b3c,color:#fff
    style Preview_Path fill:#8b6914,color:#fff
    style Migrate fill:#8b6914,color:#fff
    style Dep_Path fill:#8b1a1a,color:#fff
    style Block fill:#8b1a1a,color:#fff

Active Deadlines (as of July 29, 2026):

Foundry Workflows → retires December 1, 2026. Verbatim: “Microsoft Foundry is retiring workflows on December 1, 2026. If you’re looking to build new workflows, use Microsoft Agent Framework.” Paths: Microsoft Agent Framework (recommended) · Azure Logic Apps (visual designer) · A2A (lightweight handoffs)
Copilot Studio for Teams app → after end of June 2026, verbatim: “it will no longer be possible to use the Copilot Studio for Teams app to create classic chatbots. The app will redirect you to the Copilot Studio web app instead.” Scope: makers on a Teams plan, who are “limited to creating agents that use classic orchestration… and they can only publish these agents to Microsoft Teams.” Standalone Copilot Studio subscriptions are unaffected
Azure Cache for Redis Enterprise + Enterprise Flash → retire 2027-03-31 (disabled 2027-04-01); Basic/Standard/Premium retire 2028-09-30 (disabled 2028-10-01)
azure-ai-inference SDK → retires May 30, 2026 (successor: openai SDK)
Bot Framework → retired Dec 31, 2025 (successor: M365 Agents SDK)
Assistants API → a sunset is documented, but ⚠️ the exact date is unconfirmed against current Foundry documentation. Plan the migration to the Responses API and Foundry Agents Service; verify the date yourself before you build a program plan around it

IQ Layer Selection

Interactive version available. Pan, zoom, and click nodes below, or open the full explorer.

View as static Mermaid diagram

%%{init: {'theme':'dark'}}%%
flowchart TD
    Start([What knowledge<br/>does the agent need?]) --> Domain{Primary data domain?}

    Domain -->|Enterprise docs,<br/>files, blobs, indexes| FIQ[Foundry IQ<br/>GA via 2026-04-01 REST API<br/>portal <i>Preview</i>]
    Domain -->|M365 collaboration<br/>emails, meetings, chats| WIQ[Work IQ<br/>APIs GA 2026-06-16<br/>MCP <i>Preview</i>]
    Domain -->|Analytics, semantic<br/>models, dashboards| FBIQ[Fabric IQ<br/><i>Preview</i>]
    Domain -->|Public web,<br/>real-time information| WEBIQ[Web IQ<br/><i>Limited Access</i>]
    Domain -->|Multiple domains| Multi[Combine IQ layers]

    FIQ --> FIQ_Detail[Azure AI Search underneath<br/>knowledge bases via MCP<br/>ACL + Purview labels]
    WIQ --> WIQ_Detail[Chat, Context, Tools and<br/>Workspaces API domains<br/>Entra delegated auth only]
    FBIQ --> FBIQ_Detail[Ontologies, graph models,<br/>semantic models,<br/>operations agents]
    WEBIQ --> WEBIQ_Detail[Grounding with Bing<br/>real-time web search,<br/>custom domain scoping]

    Multi --> Combine[Each IQ layer is<br/>standalone. Agents<br/>can combine any subset]

    Combine --> FIQ_Detail
    Combine --> WIQ_Detail
    Combine --> FBIQ_Detail
    Combine --> WEBIQ_Detail

    FIQ_Detail --> License1[Azure consumption]
    WIQ_Detail --> License2[Billed in Copilot Credits<br/>access independent of<br/>M365 Copilot licensing]
    FBIQ_Detail --> License3[Fabric capacity<br/>data agent needs F2+ or P1+]
    WEBIQ_Detail --> License4[Grounding API consumption;<br/>Limited Access application required]

    style FIQ fill:#264f78,color:#fff
    style WIQ fill:#4a3278,color:#fff
    style FBIQ fill:#783228,color:#fff
    style WEBIQ fill:#0e5a6f,color:#fff
    style Multi fill:#1a6b3c,color:#fff
    style Combine fill:#1a6b3c,color:#fff

Key distinction: These four Microsoft IQ capabilities are not interchangeable. Each grounds agents in a different data domain: enterprise documents (Foundry IQ), collaboration signals (Work IQ), analytics and business logic (Fabric IQ), or live web information (Web IQ). Choose based on data domain, not brand preference.

Validation Summary: IQ Layer Selection

Last Validated: July 29, 2026

Layer	Status: read the qualifier, not the brand	What it actually is
Foundry IQ	Agentic retrieval is GA via the 2026-04-01 REST API; portal experiences remain preview-only	Knowledge bases over Azure AI Search, which was not renamed in the Foundry rebrand. The 2026-05-01-preview API adds remote sources: MCP Server, Work IQ, Fabric data agent, Fabric ontology
Work IQ	APIs GA 2026-06-16. ⚠️ Work IQ MCP is Preview	Four API domains: Chat, Context, Tools, Workspaces. Billed in Copilot Credits. Entra delegated auth only; app-only is NOT supported. Workspaces use SharePoint Embedded storage. Access is independent of M365 Copilot licensing
Fabric IQ	Preview workload	Ontologies, graph models, semantic models, and operations agents. Distinct from the Fabric data agent, which is GA and requires F2+ or P1+ capacity
Web IQ	Limited Access, requires a separate application	Grounding with Bing, real-time web search, custom domain scoping

Sources: IQ Layer Selection

Azure AI Search agentic retrieval (GA via 2026-04-01 REST API; portal preview-only)
Work IQ APIs announcement (APIs GA 2026-06-16) and Work IQ MCP overview (preview)
Fabric IQ (Preview) and Fabric data agent (GA)

Last Updated: July 29, 2026 Next: Evaluation Criteria - Score complexity, skills, budget, and governance after selecting a path

Next Steps

Detailed comparisons: → Feature Comparison

Real-world examples: → Scenarios

Evaluate readiness: → Evaluation Criteria

Visual Framework

Table of contents

Diagram Index

Complete Decision Flow

Validation Summary

UI-Based Agents (GA unless noted)

Autonomous Agents

API/Headless Services (GA)

Vector Databases

Sources: Complete Decision Flow

Persona-Based Flow

Validation Summary - Persona-Based Flow

End User (GA)

Business Maker (GA)

Developer (GA unless noted)

Developer: The Codebase Loop (Bucket 3)

Business Builder: The Natural-Language Build Path (Bucket 3)

Data Scientist/Analyst

Integration Specialist

Sources: Persona-Based Flow

Data Grounding Decision

Validation Summary - Data Grounding Decision

M365 Data Sources (GA)

Document Processing - File Search (GA)

Document Processing - Production Scale (GA)

Document Processing - Multimodal (Preview)

Structured Databases - Vector Search

Analytics Platform (GA with Preview features)

Sources: Data Grounding Decision

MCP Integration (Preview)

When to Use File Search vs Azure AI Search

Complexity Assessment Flow

Budget & Timeline Tradeoffs

Validation Summary - Budget & Timeline Tradeoffs

M365-Only ($0 AI infrastructure add’l)

Starter ($200-500/mo)

Growth ($1K-5K/mo)

Enterprise ($5K+/mo)

Timeline Estimates

Cross-Tier: The Two Prepurchase Plans

Cost Calculation Notes

Sources

Governance & Compliance Path

Validation Summary: Governance & Compliance Path

M365 Tenant Boundary Technologies

Azure Region Control Technologies

Flexible Residency Technologies

Multi-Agent Orchestration

Validation Summary: Multi-Agent Orchestration

Connected Agents / Sub-Agent Pattern

Agent Workflow Orchestration

Event-Driven Agents (Single Agent Triggering)

Sources: Multi-Agent Orchestration

Upgrade Paths

Validation Summary: Upgrade Paths

Progressive Enhancement Mechanisms

Using This Guide

Lifecycle Check

IQ Layer Selection

Validation Summary: IQ Layer Selection

Sources: IQ Layer Selection

Next Steps