This article was created using an automated generation flow powered by generative AI.
What is Copilot Studio for Teams? Creating Limited Agents in M365 and the Boundary with the Standalone Version
About This Article
Based on Microsoft's official information as of September 18, 2026, this article clarifies the boundaries between Copilot Studio use rights for Teams and standalone Copilot Studio in a way that is easy for beginners to understand.Verification Status: 📘 Microsoft Official Information Confirmed / Not Verified on Actual Environment
Copilot Studio is a Microsoft service for creating conversationalagents (mechanisms that answer questions and perform tasks). However, "Copilot Studio included in Microsoft 365" and "standalone Copilot Studio (contracted separately)" do not share the same functional scope.
The Teams-targeted plans in eligible Microsoft 365 subscriptions are basically limited tolimited use where you publish only to Teams using classic orchestration (a method of controlling conversations based on pre-configured topics and conditions). Advanced orchestration based on generative AI or publishing to channels other than Teams may require a standalone subscription.
- Key Terms to Learn First
- Differences Between the Plan for Teams and the Standalone Version
- Where is the Boundary of "Within Teams"?
- Practical Example: Building a Small General Affairs FAQ
- Visualize the Conversation Flow
- Users, Administrative Staff, and IT Admins Look at Different Things
- Limits and Quotas Are Also Not "The Same as the Standalone Version"
- Items for Administrators to Confirm Before Deployment
- Common Misconceptions
- Microsoft Official Information
Key Terms to Learn First
| Term | Meaning for Beginners |
|---|---|
| agent | A mechanism that receives user questions and responds according to configured knowledge and processes. |
| topic | A unit that defines a conversation flow, such as "if this question, then this flow." |
| classic orchestration | A method of routing conversations centered around topics prepared by the creator. |
| generative orchestration | A method where generative AI determines available knowledge, tools, etc., to construct the process. |
| Dataverse for Teams | A data platform tied to Teams, used by Power Platform apps for Teams. |
| connector | Components that connect to Microsoft 365 or external services. |
Differences Between the Plan for Teams and the Standalone Version
According to Microsoft's current licensing documentation, to create and manage agents with the plan for Teams, you need a license that allows you to use Teams and the Copilot Studio use rights for Teams.
| Aspect | Plan for Teams | Standalone Copilot Studio |
|---|---|---|
| Primary Use Case | Limited agents within Teams | Full-scale business agents |
| Orchestration | Primarily classic | Generative orchestration available |
| Publishing Destination | Teams | Multiple channels supported by Copilot Studio |
| Environment | Dataverse for Teams | Various supported Power Platform environments |
| Premium connector | Out of basic scope | Available depending on contract and functional requirements |
| Skills | Not included in the Teams plan | Available in standalone |
Instead of assuming "I have E3/E5, so I can use all features of Copilot Studio,"whether your contracted Microsoft 365 SKU includes Copilot Studio use rights for Teams, and furthermore, whether the agent you want to build fits within the scope of Teamsit is safer to verify in these two stages.
Where is the Boundary of "Within Teams"?
flowchart LR
U[Teams利用者] --> T[Microsoft Teams]
T --> A[Copilot Studio agent]
A --> O[classic orchestration]
A --> D[Dataverse for Teams]
O --> Q[FAQ / topic]
A -. 高度な機能が必要 .-> S[Standalone Copilot Studio]
S --> G[Generative orchestration]
S --> C[Teams以外のチャネル]
S --> P[Premium connectors等]
What is important in this diagram is not to think of the use rights for Teams as a "free version of the standalone product."the right to try out agents on a small scale within the boundary of TeamsThinking of it this way makes it much easier to understand.
Practical Example: Building a Small General Affairs FAQ
For example, suppose a general affairs representative wants to test an agent that guides users through only three questions: "expense reimbursement deadline," "meeting room reservation," and "work-from-home application."
Instead of loading the entire internal regulations or operating external services from the start, you decide on known questions and expected answers.
目的: 総務FAQの動作確認 質問: 経費精算の締日は? 期待する回答: 登録した社内ルールの範囲だけで回答する 不明な場合: 推測せず、総務窓口へ確認するよう案内する
What to Look For
Can the agent be called from Teams?
Does it answer according to the registered FAQ?
Does it avoid making up answers for unregistered policies?
Can only the intended users access it?
Change One Thing
Change only one item, "expense reimbursement deadline," and retest. If only the changed information is reflected in the response, you can understand the flow of update -> verification. Do not change the publishing scope or external channels at the same time in the first experiment.
Visualize the Conversation Flow
sequenceDiagram
participant U as 利用者
participant T as Teams
participant A as Agent
participant Q as FAQ/Topic
U->>T: 経費精算の締日は?
T->>A: 質問を渡す
A->>Q: 該当するtopicを確認
Q-->>A: 登録済み回答
A-->>T: 回答
T-->>U: Teams内に表示
With this small verification, it becomes easier to track what information the agent is basing its answers on.
Users, Administrative Staff, and IT Admins Look at Different Things
| Role | Key Things to Confirm |
|---|---|
| General Users | Where to use it in Teams, and whether the answers are reliable |
| Administrative Staff / Agent Creators | How to update FAQs, how to correct incorrect answers, and who to publish to |
| IT Admins / Administrators | Licenses, environments, DLP, connectors, data sources, publishing scope, trials |
DLP (Data Loss Prevention)is an important administrative feature on the Power Platform side to control which connectors can be combined and prevent unintended data exfiltration.
Limits and Quotas Are Also Not "The Same as the Standalone Version"
In Microsoft's Quotas and limits, the Teams app side has limits such as50 agents per team, and 250 topics per agent in Dataverse for Teams environments. Additionally, using skills in the Teams plan is said to require standalone Copilot Studio.
Since limits may change in the future, please recheck the official page when designing.
Items for Administrators to Confirm Before Deployment
Is the user's SKU eligible for Teams?
Is it covered by the Copilot Studio Teams plan?
Where are the Teams and Dataverse for Teams environments where the agent will be placed?
Is publishing to channels other than Teams required?
Are generative orchestration or premium connectors required?
When calling Power Automate, are the use rights and request limits appropriate?
Are the connectors planned for use allowed by the DLP policy?
Is it acceptable to include confidential information in the data referenced by the agent?
Do you allow self-signup for trials by users?
Common Misconceptions
"If I have Microsoft 365 E3/E5, the standalone version of Copilot Studio is also included at no additional cost."
Not necessarily. You need to think of the Teams use rights on the Microsoft 365 side and standalone Copilot Studio separately.
"If I can build an agent in Teams, I can publish it directly to a website as well."
The plan for Teams has restrictions on publishing destinations. If you need channels other than Teams, recheck the licensing and functional requirements.
"Since it has 'Copilot' in the name, it is the same as the Microsoft 365 Copilot license."
Microsoft 365 Copilot and Copilot Studio have different roles and licenses. It is important not to judge use rights based on the name alone.
Microsoft Official Information
Copilot Studio for Teams is convenient for starting with FAQs and small-scale agents within Teams, butwhere to publish, which data to connect to, and how much of the generative AI features to useonce these expand, it is safer to re-verify the boundary with standalone Copilot Studio.
