diff --git a/copilot-agent-strategy/copilot-agents-guide/README.md b/copilot-agent-strategy/copilot-agents-guide/README.md
index e8b752c..f239591 100644
--- a/copilot-agent-strategy/copilot-agents-guide/README.md
+++ b/copilot-agent-strategy/copilot-agents-guide/README.md
@@ -6,9 +6,9 @@ summary: >-
Compare Microsoft Copilot agent platforms with capability views, decision guidance, support
indicators, and scenarios.
author: Microsoft FastTrack
-version: 3.0.0
+version: 3.1.0
published: "2025-10-28"
-updated: "2026-04-17"
+updated: "2026-07-31"
tags:
- guide
- decision
@@ -76,15 +76,15 @@ The **Copilot Agents Guide** is a comprehensive, interactive web-based tool that
- Key takeaways
### β‘ FastTrack Support Indicators
-- Subtle badges showing which agents qualify for Microsoft FastTrack deployment support
+- Subtle badges showing which agents qualify for Microsoft FastTrack remote guidance
- Link to official FastTrack service description
-- 5 of 6 agent types include FastTrack support
+- 5 of 6 agent types include FastTrack remote guidance
## π€ Agent Types Covered
### 1. **Researcher & Analyst Agents**
Built-in reasoning agents for research and data analysis
-- **Researcher:** Complex multi-step research with multi-model Critique (generation + evaluation) and Model Council (side-by-side model comparison)
+- **Researcher:** Complex multi-step research with a model picker for GPT and Claude. Critique (a second reasoning pass by Claude) and Model Council (parallel GPT and Claude comparison) require the Frontier program
- **Analyst:** Data analysis with Python execution
- **Availability:** Immediate (pre-pinned in M365 Copilot)
- **Cost:** Included with M365 Copilot license (25 queries/month)
@@ -152,7 +152,7 @@ Upload the HTML file to any web server or hosting platform:
1. **Start with Overview** - Understand each agent type's value proposition
2. **Check Guidance Tab** - Review the 4-step decision framework
3. **Compare Costs** - See time-to-market and licensing requirements
-4. **Consider FastTrack** - Note which agents include deployment support
+4. **Consider FastTrack** - Note which agents include remote guidance
### For IT Architects
1. **Review Comparison Tab** - Analyze technical capabilities side-by-side
@@ -213,7 +213,7 @@ All agent information verified from official Microsoft sources:
- FastTrack for Microsoft 365 service descriptions
- Official Microsoft 365 Blog announcements
-**Last Verified:** April 2026
+**Last Verified:** July 2026
## π Updates & Maintenance
@@ -227,8 +227,9 @@ This guide is updated to reflect:
## β οΈ Important Notes
-### FastTrack Support
-- 5 of 6 agent types qualify for FastTrack deployment assistance
+### FastTrack Remote Guidance
+- 5 of 6 agent types qualify for FastTrack remote guidance
+- Custom engine agents are covered only when the deployment channel is Teams, Microsoft 365 Copilot, or SharePoint
- Microsoft 365 Agents Toolkit (pro-code) is self-service only
- See footer link for detailed FastTrack service description
@@ -271,8 +272,8 @@ This guide is provided as-is for informational and planning purposes. Microsoft,
---
-**Version:** 3.0 (April 2026)
-**Includes:** Researcher Critique & Model Council, Copilot Cowork (Frontier), Agent 365 GA, Wave 3 updates, E7 Frontier Suite
+**Version:** 3.1 (July 2026)
+**Includes:** Researcher Critique & Model Council (Frontier), Copilot Cowork GA, Agent 365 GA, Wave 3 updates, E7 Frontier Suite
**Format:** Single-file HTML application
*Built to help you navigate the Microsoft Copilot agent ecosystem with confidence.*
\ No newline at end of file
diff --git a/copilot-agent-strategy/copilot-agents-guide/index.html b/copilot-agent-strategy/copilot-agents-guide/index.html
index 25c0ea9..d19ecb3 100644
--- a/copilot-agent-strategy/copilot-agents-guide/index.html
+++ b/copilot-agent-strategy/copilot-agents-guide/index.html
@@ -248,7 +248,7 @@
technicalLevel: 'Beginner',
shortName: 'First-Party Agents',
fastTrackSupported: true,
- description: 'Ready-to-use agents built into Microsoft 365 Copilot that deliver immediate value without custom development. Powered by multi-model intelligence from OpenAI and Anthropic, with Researcher Critique and Model Council for enterprise-grade research. From deep research and data analysis to community knowledge sharing and workflow automation, these agents cover the most common productivity scenarios out of the box.',
+ description: 'Ready-to-use agents built into Microsoft 365 Copilot that deliver immediate value without custom development. Powered by multi-model intelligence from OpenAI and Anthropic, with Researcher Critique and Model Council available to Frontier program participants. From deep research and data analysis to community knowledge sharing and workflow automation, these agents cover the most common productivity scenarios out of the box.',
subAgents: [
{
id: 'researcher',
@@ -259,10 +259,10 @@
borderColor: 'border-indigo-200 dark:border-indigo-800/30',
status: 'GA',
statusColor: 'bg-green-100 text-green-800 dark:bg-green-900/30 dark:text-green-300',
- description: 'Complex multi-step research powered by multi-model intelligence. Critique separates generation from evaluation using models from Anthropic and OpenAI for higher accuracy on deep research tasks. Model Council lets you compare responses from different models side by side.',
+ description: 'Complex multi-step research powered by multi-model intelligence, with a model picker for GPT and Claude. Critique and Model Council are additional research modes that require the Frontier program.',
capabilities: [
- 'Critique: multi-model architecture β one model generates, another reviews and refines',
- 'Model Council: side-by-side model comparison with agreement/divergence analysis',
+ 'Critique (Frontier): a report is generated by GPT, then given a second reasoning pass by Claude',
+ 'Model Council (Frontier): runs GPT and Claude in parallel and highlights where they agree and diverge',
'Multi-source research synthesis across work files, emails, meetings, and web',
'Microsoft Graph connector support',
'Structured reports with source citations and rubric-based evaluation',
@@ -314,9 +314,9 @@
color: 'text-amber-500',
bgColor: 'bg-amber-50 dark:bg-amber-950/20',
borderColor: 'border-amber-200 dark:border-amber-800/30',
- status: 'Preview',
- statusColor: 'bg-amber-100 text-amber-800 dark:bg-amber-900/30 dark:text-amber-300',
- description: 'Automate routine tasks directly from Microsoft 365 Copilot β create flows, set reminders, send notifications, and manage approvals using natural language. Rolling out in Frontier; English-first.',
+ status: 'Available',
+ statusColor: 'bg-slate-100 text-slate-800 dark:bg-slate-800/40 dark:text-slate-300',
+ description: 'Automate routine tasks directly from Microsoft 365 Copilot β create flows, set reminders, send notifications, and manage approvals using natural language. Available within Microsoft 365 Copilot and controlled by admins from the Microsoft 365 admin center; English only.',
capabilities: [
'Natural language workflow creation',
'Pre-built templates for common automations',
@@ -331,18 +331,19 @@
'Researcher: Multi-source synthesis with Microsoft Graph connector support',
'Analyst: Advanced data analysis with Python and chain-of-thought reasoning',
'Viva Engage: Proactive community Q&A with expert verification (preview)',
- 'Workflows: Natural language task automation and approvals (Frontier preview)',
+ 'Workflows: Natural language task automation and approvals (English only)',
'All: Pre-built into Microsoft 365 Copilot β no custom development',
'All: Immediate or same-day deployment',
- 'Researcher: Multi-model intelligence with Critique and Model Council',
+ 'Researcher: Critique and Model Council research modes (Frontier program)',
'Researcher: Over 30 languages supported',
- 'Researcher: 25 queries per user per month',
'Viva Engage: Grounded in community conversations and SharePoint',
'Workflows: Integration with Power Automate templates'
],
limitations: [
'Requires Microsoft 365 Copilot license',
'Researcher: 25 queries per user per month',
+ 'Researcher: Critique and Model Council require the Frontier program',
+ 'Workflows: English only',
'No support for external deployment channels',
'Viva Engage: Requires active communities with conversation history for best results',
'Limited customization β designed for specific built-in scenarios'
@@ -371,7 +372,7 @@
{ step: 6, title: 'Explore Workflows Agent', description: 'Ask Copilot to automate a routine task β try "Create an approval flow for expense reports" to see natural language automation.' }
],
bestPractices: [
- 'Use Researcher for multi-source synthesis β Critique (default) uses multi-model review for higher accuracy, or use Model Council to compare models side by side',
+ 'Use Researcher for multi-source synthesis β if your tenant is in the Frontier program, Critique adds a second reasoning pass and Model Council compares GPT and Claude side by side',
'Use Analyst for structured data β spreadsheets, CSVs, and datasets with Python execution',
'Enable Viva Engage Community Agent in communities with active discussions and identified subject-matter experts',
'Designate community experts to verify and mark AI-drafted answers as "Trusted" for knowledge curation',
@@ -384,7 +385,7 @@
{ item: 'Viva Engage communities identified for agent enablement', critical: false },
{ item: 'Community experts designated for answer verification', critical: false },
{ item: 'Sample prompts shared with team for each agent', critical: false },
- { item: 'Usage monitoring configured for Researcher', critical: false }
+ { item: "Researcher's 25-query-per-user monthly limit communicated to users", critical: false }
]
}
},
@@ -398,21 +399,20 @@
technicalLevel: 'Beginner',
shortName: 'SharePoint',
fastTrackSupported: true,
- description: 'AI agents that live within SharePoint, now supporting cross-site querying through Copilot, Teams integration, and Knowledge Agent metadata enrichment (preview) for smarter discovery.',
+ description: 'AI agents that live within SharePoint, now supporting cross-site querying through Copilot and sharing into Microsoft Teams conversations.',
capabilities: [
'Site-specific knowledge grounding',
'Automatic content indexing',
'SharePoint permissions respect',
'Document summarization',
'Cross-site querying via Copilot integration',
- 'Teams 1:1 chat with agents',
- 'Microsoft Lists natural language queries',
- 'Knowledge Agent metadata enrichment (preview)'
+ 'Shareable into Teams group chats, meeting chats, and channels via @mention'
],
limitations: [
'Basic conversation capabilities only',
'No external integrations',
- 'Knowledge Agent metadata features still rolling out',
+ 'Up to 20 source items per agent β sites, document libraries, folders, and files combined',
+ 'Teams 1:1 chat with SharePoint-created agents is not supported yet',
'Cross-site requires M365 Copilot license'
],
whenToUse: [
@@ -433,7 +433,7 @@
gettingStarted: [
{ step: 1, title: 'Choose your SharePoint site', description: 'Identify the site with the content you want your agent to access.' },
{ step: 2, title: 'Create the agent', description: 'From the site home, a library, or a list: choose New > Agent (or use the AI actions > Create an agent option) to build your SharePoint agent.' },
- { step: 3, title: 'Configure knowledge sources', description: 'Select document libraries and lists the agent should reference.' },
+ { step: 3, title: 'Configure knowledge sources', description: 'Select the sites, document libraries, folders, and files the agent should reference β up to 20 source items in total.' },
{ step: 4, title: 'Test and share', description: 'Test the agent with sample questions, then share with your team via the site.' }
],
bestPractices: [
@@ -518,19 +518,21 @@
technicalLevel: 'Intermediate',
shortName: 'Copilot Studio Declarative',
fastTrackSupported: true,
- description: 'Declarative agents built in Copilot Studio Full that extend Microsoft 365 Copilot using its orchestrator, with Agent Store publishing and support for Copilot Studio connected-agent patterns (preview) for advanced scenarios.',
+ description: 'Declarative agents built in Copilot Studio Full that extend Microsoft 365 Copilot using its orchestrator, with Agent Store publishing and support for connecting to other agents for advanced scenarios.',
capabilities: [
'Uses M365 Copilot orchestrator',
'Custom instructions and behavior',
'Enterprise knowledge integration',
'Plugin actions for external systems',
'Agent Store publishing and distribution',
- 'Copilot Studio connected-agent scenarios (preview), including external A2A connections',
+ 'Classic experience: external agent connections over the A2A protocol (generally available)',
+ 'New agent experience: connected agents (preview) β currently limited to other Copilot Studio agents',
'Available in Microsoft 365 Copilot and Microsoft 365 apps (Teams, Word, Excel, Outlook)',
'Inherits M365 Copilot security'
],
limitations: [
- 'Users need Microsoft 365 Copilot Chat (included with most M365 subscriptions) β full M365 Copilot license is optional; usage billed via Copilot Credits, or free for M365 Copilotβlicensed users',
+ "Access and billing depend on the agent's grounding and capabilities: agents grounded only in instructions or public web data can be used without pay-as-you-go, while SharePoint, file, Dataverse, and Copilot connector grounding requires pay-as-you-go or a Microsoft 365 Copilot license, and email, people, and Teams grounding requires a Microsoft 365 Copilot license. Eligible licensed usage is included, subject to fair-use limits.",
+ 'Agents created in the new agent experience cannot be converted to the classic experience, and vice versa',
'Limited to Microsoft 365 ecosystem',
'Plugin development requires technical skills',
'RAI validation checks required for publishing'
@@ -562,7 +564,7 @@
'Test with the M365 Copilot orchestrator context in mind β behavior differs from standalone testing'
],
deploymentChecklist: [
- { item: 'Copilot Chat (included with most M365 subscriptions) or M365 Copilot licenses for target users; Copilot Credits configured for Chat users', critical: true },
+ { item: 'Microsoft 365 Copilot Chat available to target users; pay-as-you-go or Copilot Credits configured where the agent grounds in tenant data', critical: true },
{ item: 'Copilot Studio environment provisioned', critical: true },
{ item: 'Agent instructions and knowledge sources defined', critical: true },
{ item: 'Plugin actions configured (if needed)', critical: false },
@@ -580,17 +582,17 @@
technicalLevel: 'Intermediate',
shortName: 'Copilot Studio Custom Engine',
fastTrackSupported: true,
- description: 'Full-featured Copilot Studio with advanced capabilities, including multi-agent orchestration, Agent Store publishing, Computer Use automation for legacy apps (preview), GPT-5 Chat availability, custom connectors, multi-channel publishing, and enterprise-grade features.',
+ description: 'Full-featured Copilot Studio with advanced capabilities, including multi-agent orchestration, Agent Store publishing, computer use for browser and Windows desktop automation, a choice of supported OpenAI and Anthropic models, custom connectors, multi-channel publishing, and enterprise-grade features.',
capabilities: [
'Advanced conversation flows and logic',
'Custom connectors and API integrations',
'Multi-channel publishing',
'Multi-agent orchestration (agent collaboration)',
'Agent Store marketplace publishing',
- 'Computer Use automation for legacy apps (preview)',
+ 'Computer use for browser and Windows desktop UI automation (generally available)',
'Human-in-the-loop (HITL) controls',
'Automated agent evaluation and testing',
- 'GPT-5 Chat model support',
+ 'Choice of supported OpenAI and Anthropic models; availability varies by region and agent experience',
'Power Platform integration',
'Advanced analytics and monitoring',
'Custom authentication and security'
@@ -598,6 +600,8 @@
limitations: [
'Requires technical expertise and training',
'Copilot Studio license required',
+ 'Agents created in the new agent experience cannot be converted to the classic experience, and vice versa',
+ 'FastTrack remote guidance covers custom engine agents only when the deployment channel is Teams, Microsoft 365 Copilot, or SharePoint',
'Higher development time investment required',
'Ongoing maintenance and monitoring needed'
],
@@ -606,7 +610,7 @@
'Customer service and support at scale',
'Multi-step workflows and approval processes',
'Multi-agent workflows requiring orchestration',
- 'Legacy app automation without APIs',
+ 'Browser and Windows desktop automation for apps without APIs',
'Integration with external systems and APIs'
],
metrics: {
@@ -628,7 +632,7 @@
'Use the evaluation and testing framework to maintain quality as you scale',
'Implement human-in-the-loop for high-stakes decisions',
'Monitor analytics closely in the first 30 days to catch issues early',
- 'Plan your message credit budget based on expected conversation volume'
+ 'Forecast Copilot Credit consumption based on agent design, enabled features, and expected usage volume'
],
deploymentChecklist: [
{ item: 'Copilot Studio license procured', critical: true },
@@ -651,15 +655,15 @@
technicalLevel: 'Expert',
shortName: 'M365 Agents Toolkit',
fastTrackSupported: false,
- description: 'Build declarative and custom engine agents with full developer control using VS Code, Visual Studio, and CLI extensions, with Agent Store publishing, Microsoft Teams SDK, and MCP protocol support (A2A via Teams SDK). Note: Pro-code development is outside FastTrack deployment scope.',
+ description: 'Build declarative and custom engine agents with full developer control using VS Code, Visual Studio, and CLI extensions, with Agent Store publishing, Microsoft 365 Agents SDK integration, and support for adding remote MCP servers as plugins. Note: FastTrack does not cover creating custom agents with developer toolchains.',
capabilities: [
'Full pro-code development workflow',
- 'Microsoft Teams SDK (formerly Teams AI Library)',
+ 'Microsoft 365 Agents SDK integration for multichannel custom engine agents',
'Direct integration with Azure services',
'Version control and CI/CD support',
'Advanced plugin development',
'Agent Store multi-channel publishing',
- 'MCP protocol support (A2A available via Teams SDK)',
+ 'Add remote MCP servers as plugins for declarative agents',
'Agents Playground for local testing',
'VS Code, Visual Studio, and CLI extensions',
'Custom engine agent support (bring your own LLM)'
@@ -669,7 +673,8 @@
'More complex setup and maintenance',
'Steeper learning curve',
'Requires DevOps practices',
- 'Not covered by FastTrack deployment assistance (pro-code is out of scope)'
+ 'A2A is implemented by the agent runtime using dedicated protocol SDKs β it is not provided by the Toolkit itself',
+ 'Creating custom agents with developer toolchains is outside FastTrack scope β see the FastTrack service description'
],
whenToUse: [
'Need full control over agent behavior',
@@ -701,7 +706,7 @@
deploymentChecklist: [
{ item: 'Development environment configured (VS Code/Visual Studio)', critical: true },
{ item: 'Azure subscription provisioned', critical: true },
- { item: 'Teams SDK and toolkit extensions installed', critical: true },
+ { item: 'Microsoft 365 Agents Toolkit extension installed for VS Code or Visual Studio', critical: true },
{ item: 'CI/CD pipeline configured', critical: false },
{ item: 'Agents Playground tested locally', critical: false },
{ item: 'Agent Store submission prepared', critical: false }
@@ -859,7 +864,7 @@ Start with Your
-
Multi-step personal tasks or delegated knowledge work? β Copilot Cowork (Available β Frontier)
+
Multi-step personal tasks or delegated knowledge work? β Copilot Cowork (Generally available)
@@ -883,7 +888,7 @@
Start with Your
-
External audience or complex workflows? β Copilot Studio custom engine agents
+
External audience or complex workflows? β Copilot Studio custom engine agents (FastTrack remote guidance covers Teams, Microsoft 365 Copilot, and SharePoint channels)
@@ -941,7 +946,7 @@
Consider Budget
- Copilot Cowork: Immediate for Frontier-licensed users (Available)
+ Copilot Cowork: Available after tenant enablement and usage-based billing configuration (Generally available)
@@ -993,7 +998,7 @@
Plan for Growth
Scale Up
- Start with Copilot Cowork and first-party agents like Researcher, Analyst, Viva Engage Community Agent, or SharePoint agents β these are the fastest way to demonstrate personal productivity value. Prove adoption with Agent Builder, then scale to Copilot Studio or the Toolkit when you need repeatable multi-user processes, compliance guardrails, governance, or external channels. Think of Cowork as the front door β and structured agents as the deeper rooms your organization will explore once inside. FastTrack can help with deployment planning, change management, and adoption at every stage.
+ Start with Copilot Cowork and first-party agents like Researcher, Analyst, Viva Engage Community Agent, or SharePoint agents β these are the fastest way to demonstrate personal productivity value. Prove adoption with Agent Builder, then scale to Copilot Studio or the Toolkit when you need repeatable multi-user processes, compliance guardrails, governance, or external channels. FastTrack provides remote guidance for deployment planning within its published scope; adoption and change management strategy are not part of the FastTrack benefit.
@@ -1008,7 +1013,7 @@
Cowork vs. Structured Agents: Choosing the Right Approach
- Copilot Cowork is now available for Frontier companies, giving customers clear guidance on when to use adaptive AI delegation vs. structured agent automation.
+ Copilot Cowork is generally available to Microsoft 365 Copilot tenants worldwide in tier-1 languages, giving customers clear guidance on when to use adaptive AI delegation vs. structured agent automation.
@@ -1064,28 +1069,6 @@
Structured β Agent Bui
-
-
-
-
The Composability Advantage
-
-
Cowork and structured agents are not mutually exclusive β they are composable. Copilot Studio agents can be invoked within a Cowork plan, enabling sophisticated workflows where:
-
-
-
Cowork
-
Handles open-ended orchestration and reasoning
-
-
-
-
Studio Agent
-
Handles structured subprocesses within the plan
-
-
-
Don't think of it as "pick one" β combining Cowork's adaptive reasoning with Studio's structured guardrails lets you get the best of both worlds.
-
@@ -1131,7 +1114,7 @@
Immediate: First-party agents (Researcher, Analyst, Viva Engage, Workflows)
-
Immediate: Copilot Cowork (Available β Frontier)
+
Tenant setup required: Copilot Cowork (Generally available)
Minutes to hours: SharePoint agents
@@ -1157,7 +1140,7 @@
Copilot Cowork:
-
Available through the Frontier program for Microsoft 365 Copilot licensed users. Check current licensing details for Cowork availability and metered usage.
+
Generally available to Microsoft 365 Copilot tenants worldwide in tier-1 languages. Requires an active Microsoft 365 Copilot license, a supported browser, Cowork enabled, usage-based billing enabled, and Anthropic approved as a tenant subprocessor. Local browser use in Microsoft Edge remains a Frontier capability. Check current licensing details for metered usage.
Agent 365 β enterprise agent governance:
@@ -1258,23 +1241,6 @@
-
-
-
SCENARIO
-
Orchestrated workflow combining personal delegation with structured business logic
-
-
-
RECOMMENDATION
-
Cowork + Copilot Studio agent
-
-
-
WHY?
-
Cowork handles open-ended reasoning and orchestration while Studio agents execute governed subprocesses β the composability advantage
-
-
-
@@ -1352,8 +1318,8 @@
π Key Takeaways
-
πͺ Use Cowork as the Front Door
-
Copilot Cowork is the easiest on-ramp for individual users to see immediate value. Once engaged, they will naturally want to explore structured agents for team and business processes.
+
πͺ Cowork Is a Low-Friction Starting Point
+
Copilot Cowork lets individual users delegate multi-step work without building anything, so it can be a practical first step. Once engaged, they often want to explore structured agents for team and business processes.
Start with Built-In Value
@@ -1361,7 +1327,7 @@ Start with Buil
π§ Right Tool for the Right Job
-
Cowork excels at personal, adaptive work. Studio agents shine for repeatable, governed, multi-user processes. Help customers understand the spectrum β and the composability between them.
+
Cowork excels at personal, adaptive work. Studio agents shine for repeatable, governed, multi-user processes. Help customers understand the spectrum and pick the right tool for each scenario.
Then Scale with Agent Builder & Studio
@@ -1547,7 +1513,7 @@
{agent.timeToMarket}
{agent.fastTrackSupported && (
-
+
FastTrack Supported
@@ -1640,7 +1606,7 @@
+
FastTrack Supported
@@ -1920,7 +1886,7 @@
- | FastTrack Support |
+ FastTrack Remote Guidance |
{Object.entries(agentTypes).map(([key, agent]) => (
{agent.fastTrackSupported ?
@@ -2434,7 +2400,7 @@
|