Copilot Studio 2025 Updates – What’s New in Microsoft Copilot Studio

Copilot Studio 2025 Updates – What’s New in Microsoft Copilot Studio?

No comments

Loading

Microsoft has rolled out exciting new features in Copilot Studio in July 2025. Whether you’re a Power Platform developer, Microsoft 365 enthusiast, or a business user exploring AI, these updates bring powerful enhancements to make your custom Copilots smarter, faster, and more secure.

In this article, we will walk you through the top features added to Microsoft Copilot Studio in July 2025, explained in simple terms. Let’s dive in!

What’s New in Microsoft Copilot Studio?

Following are the new features or updates in Copilot Studio as of July 2025:

New “Tools” Menu in Copilot Studio

One of the biggest changes is the introduction of the Tools menu. This menu allows makers to:

  • Add reusable tools like APIs, Power Automate flows, or SharePoint actions

  • Manage, test, and edit tool parameters

  • Use the same tool across multiple copilots

Think of it as a central place where you store all the “actions” your Copilot can perform.

There are two tools menus in Copilot Studio; these are:

  •  Local Tools Menu, Scope to Particular Agent
  • Global Tools Menu, Scope to All Agents in an Environment
Tools menu in Copilot Studio
Tools menu in Copilot Studio

Let’s understand what the tools menu offers.

What are tools in Copilot Studio?

In Copilot Studio, Tools are reusable actions or functions that your Copilot can perform—like calling a Power Automate flow, accessing a SharePoint list, or sending data to an external API. You can add, manage, and test these tools in the new “Tools” menu, and use them across multiple copilots to automate tasks and enhance conversations intelligently.

Example:

Let’s say you’re building a Copilot for an HR team.

  • You create a Power Automate flow that checks the leave balance of an employee from a SharePoint list.
  • You then add this flow as a Tool in Copilot Studio.
  • In your Copilot topic, you define a question like:
  • “How many leave days do I have left?”
  • When the user asks this, Copilot automatically calls the “Check Leave Balance” tool, sends the user ID, and fetches the result.
  • The Copilot replies:
    “You have 6 casual leave days and 4 sick leave days remaining.”

Why Tools Are Useful:

  • You can reuse the same tool across different copilots

  • You don’t need to write backend logic repeatedly

  • You can test and debug each tool directly in the Tools menu

  • Supports both Power Automate and custom API endpoints

Note:

  • You can create your custom tool and add it to the tools library, or there are plenty of out-of-the-box tools available in Copilot Studio. Those are nothing but various actions from a particular connector in Power Automate flow. Now, with the help of the tool, you don’t have to create Power Automate flow explicitly; you can add a tool and use it in your flow.
Example of tools in Copilot Studio
Example of tools in Copilot Studio

I will write a separate article on how to work with tools in Copilot Studio. Now I will focus on the rest of the updates in Copilot Studio.

GPT-4o and GPT-4.1 Model Support

With this update, Microsoft Copilot Studio now supports GPT-4o and GPT-4.1 models. You can choose which model your copilot should use for generating responses.

  • GPT-4o is faster and supports multi-modal inputs

  • GPT-4.1 provides deeper reasoning and text understanding

You can configure the response model from the “Generative AI” “Responses” section, as shown below in the image:

Configure response model in Copilot Studio
Configure response model in Copilot Studio

Model Context Protocol (MCP)

The Model Context Protocol (MCP) is a new way to connect your Copilot Studio with external servers, APIs, or knowledge systems securely.

  • MCP allows real-time syncing of tools from remote servers

  • If tools are updated at the server end, changes reflect automatically in Copilot Studio

  • Ensures enterprise-level security and scalability

To get into the Model Context Protocol (MCP), you need to follow the below navigation:

Tools menu -> Add a Tool -> Model Context Protocol

Model Context protocol in Copilot Studio
Model Context protocol in Copilot Studio

Deep Reasoning Capability

Microsoft has added Deep Reasoning options that help copilots generate more logical, context-aware, and detailed responses.

When enabled, your Copilot can:

  • Understand complex queries

  • Refer to multiple data points before replying

  • Generate human-like explanations

You can enable or disable the deep reasoning feature from your agent’s Generative AI section.

Deep reasoning configuration in Copilot Studio
Deep reasoning configuration in Copilot Studio

Knowledge Section Enhancements

The Knowledge tab now lets you connect to multiple knowledge sources, including:

  • Websites

  • Dataverse

  • SharePoint document libraries

  • OneDrive folders

This helps your Copilot give richer, fact-based answers using your own data.

You can enable or disable the use of general knowledge and use information from the web using your agent’s generative knowledge section as shown in the below image.

Knowledge configuration in Copilot
Knowledge configuration in Copilot

Know about use of General Knowledge:

The foundational knowledge that the generative AI was trained on. To ground your agent only with your specific knowledge sources, turn this off.

Know about use information from the web:

Let your agent browse the Web using Bing Web search.

Search – Tenant Graph Grounding with Semantic Search

A major upgrade in this release is the integration of Tenant Graph Grounding with Semantic Search.

Here’s what it means in simple terms:

  • Your Copilot can now search across your Microsoft 365 tenant (like SharePoint, Teams, OneDrive, Outlook, and more)

  • It uses semantic search, which understands the meaning behind user queries—not just keywords

  • The results are grounded in your organization’s Graph data (Microsoft Graph), so the answers are more accurate and personalized

  • Helps your Copilot retrieve answers that are contextually correct and relevant to your tenant

This makes your AI assistant truly intelligent—it understands your question, searches across your org’s content, and brings the most useful answer back.

You can configure the tenant graph grounding with semantic search from your agent’s generative AI search section, as shown in the below image.

Tenant graph grounding with semantic search
Tenant graph grounding with semantic search

Content Moderation Settings

You can now enable or disable content moderation for your Copilot. This feature is useful for:

  • Preventing harmful or offensive outputs

  • Complying with organizational safety guidelines

  • Customizing tone and output length

You can configure your agent’s content moderation from the Generative AI moderation section as shown in the below image:

Content moderation level in Copilot Agent
Content moderation level in Copilot Agent

Content moderation level:

Lower moderation increases the risk of harmful content in your agent’s responses. Higher moderation lowers that risk, but may reduce the number of responses.

Connected Agents & AI Orchestration

This update introduces connected agents, meaning multiple Copilots can work together and orchestrate responses using a unified flow. It’s like giving your agents teamwork skills!

You can also define how the Copilot should:

  • Search across knowledge sources

  • Combine generative AI + search for accurate results

You can configure the connected agents setting from your agent’s Generative AI connected agents section as shown in the below image:

Connected agents configuration in Copilot Studio
Connected agents configuration in Copilot Studio

Connected agents:

Let other agents connect to and use this one – Let agents work together to complete workflows.

File Processing & Response Formatting

Copilot Studio now supports:

  • Custom response formatting

  • Image uploading and file processing

  • Tool configuration with input/output validation

This is especially useful when integrating Copilot with document management systems.

You can enable or disable the file upload mechanisms for your agent from the Generative AI File Processing Capabilities section as shown in the below image:

File processing capabilities - Image Uploads settings
File processing capabilities – Image Uploads settings

Image uploads:

Users can upload images (.png, .webp, .jpeg and non-animated .gif) that agents can use in conversations. Uploads are limited to 15 MB.

SharePoint Actions as Tools

You can now add SharePoint actions (like list updates, document reads, etc.) as tools in your Copilot. This is extremely useful for:

  • Automating SharePoint workflows

  • Fetching data from SharePoint lists/libraries

  • Triggering Power Automate flows via Copilot

The below image displays the example of various tools from the SharePoint connector:

SharePoint Actions as Tools in Copilot Studio
SharePoint Actions as Tools in Copilot Studio

Summary Table of Key Updates

Feature Description
Tools Menu Manage, add, and reuse actions/tools across agents
GPT-4o & GPT-4.1 Choose smarter response models with better speed and accuracy
Model Context Protocol (MCP) Securely connect Copilot Studio with external servers and tools
Deep Reasoning Enable advanced logical thinking in AI responses
Knowledge Sources Add SharePoint, OneDrive, and web data to boost knowledge-based answers
Tenant Graph Grounding & Search AI searches your tenant with semantic understanding for better results
Content Moderation Control and customize AI behavior and safety filters
Connected Agents & Orchestration Link multiple Copilots together for team-based AI interactions
Response Formatting & Files Customize output format and enable file uploads
SharePoint Actions as Tools Use SharePoint workflows directly from within Copilot Studio

YouTube Video: What’s New in Microsoft Copilot Studio?

Final Thoughts

The Copilot Studio July 2025 updates bring a mix of intelligence, connectivity, and security improvements. Features like the new Tools menu, Model Context Protocol, Tenant Graph Grounding, and Deep Reasoning make Copilot Studio more capable than ever before.

Whether you’re creating internal support bots, automating business processes, or integrating SharePoint with AI—these features will help you build smarter copilots with less effort.

 

About Post Author

Do you have a better solution or question on this topic? Please leave a comment