Skip to content
Home » How to Use Custom Connectors in Microsoft Copilot Studio — No Power Automate

How to Use Custom Connectors in Microsoft Copilot Studio — No Power Automate

Are you looking to make your Microsoft Copilot smarter by integrating custom APIs or external services? With the latest updates to Microsoft Copilot Studio, you can now connect custom connectors directly—no Power Automate flow required. In this guide, we’ll walk you through how to use your custom connector within a Copilot topic, gather variables using AI-driven topic inputs, and deliver a smoother user experience.

If you’re building AI agents to automate tasks, process orders, or fetch information from third-party systems, this post is for you. Let’s dive into how to unlock this powerful feature!

Why Use Custom Connectors in Copilot Studio?

Microsoft Copilot Studio was initially built to access external data via Power Automate flows, but this often added extra steps and complexity. Now, with native custom connector integration in preview, you can:

  • Call external APIs directly from a Copilot topic
  • Reduce latency by eliminating Power Automate dependencies
  • Simplify your logic and maintainability
  • Use variables and topic inputs more efficiently

Whether you’re building a support bot or automating internal operations, direct access to custom connectors can make your Copilot much more powerful.

What Can You Do with Custom Connectors?

Once connected, your Copilot agent can interact with custom APIs as part of the conversation. Here are a few practical use cases:

  • Process Online Orders: Let users place coffee orders, food deliveries, or service bookings directly from chat
  • Fetch External Data: Retrieve data from external CRMs, ERPs, or proprietary databases
  • Trigger Business Processes: Start workflows based on user input—without building a full Power Automate flow

The possibilities are endless and now easier to implement.

Step-by-Step: How to Use a Custom Connector in Copilot Studio

1. Set Up Your Topic in Copilot Studio

Start by navigating to your Copilot in Copilot Studio, and open or create a new Topic. In our example, we’re working with a “Coffee Copilot” designed to process customer orders.

2. Add an Action Node with Your Custom Connector

Once in the topic editor, select the “+ Add Node” button and choose “Add an action.”
You’ll notice several tabs—be sure to check the “Connector (preview)” tab. This is where your available custom connectors will appear. If your connector has already been created in Power Platform, it should show up here.

✅ Note: This feature is still in preview, so availability may vary by environment.

3. Define Connector Inputs Using Variables

Your connector likely has required inputs—these are parameters that need to be supplied by the user (e.g., coffee type, size, name).

There are two ways to capture this input:

  • Hardcoded Questions on the Canvas
  • Topic Inputs (Recommended)

Let’s explore both.

Option 1: Using Hardcoded Questions (Less Recommended)

You can add question nodes above your action and assign their responses to variables. For example:

  • “What type of coffee would you like?”
  • “What size?”
  • “What’s your name?”

You then map these variables to your connector’s inputs.

But here’s the catch: this method can make your Copilot feel robotic and repetitive, especially if the user already provides all the info in a single message.

Option 2: Use Topic Inputs for a Smarter Experience (Recommended)

Instead of hardcoding, use Topic Inputs under the Topic Details tab. These inputs allow your Copilot to intelligently extract values using generative AI.

💡 For example, if a user says: “Hi, I’m Griffin and I’d like a small black coffee,” Copilot can extract name=Griffin, size=small, coffee=black without prompting three separate times.

How to Add Topic Inputs:

  1. Go to your topic
  2. Click Details
  3. Navigate to the “Topic Inputs” tab
  4. Define each expected variable (e.g., coffeeType, coffeeSize, userName)

Then, map these inputs to the connector’s parameters in the action node.

Real-World Example: Ordering a Coffee

Let’s say your custom connector requires:

  • coffeeType
  • size
  • customerName

By using topic inputs, your Copilot can dynamically extract these values from a natural message like:

“I’d like to order a small vanilla latte. My name is Taylor.”

And send them directly to the connector—no back-and-forth needed.

Bonus: Use Dataverse as a Knowledge Source (Preview)

Another exciting preview feature is the ability to add Dataverse as a knowledge source. This allows your Copilot to pull contextual answers from business data tables—perfect for internal support bots or customer service agents.

📺 Want to learn more about connecting Dataverse? Check out our full walkthrough [insert video link].

Tips for Getting the Most Out of Custom Connectors

  • Test in Dev First: Preview features can behave differently depending on your environment
  • Use Descriptive Topic Inputs: Clear naming helps both humans and AI understand the flow
  • Optimize the Experience: Lean on AI-powered inputs instead of rigid form-filling prompts
  • Secure Your APIs: Make sure only trusted Copilots and users can access sensitive endpoints

In Conclusion

Integrating custom connectors directly into Microsoft Copilot Studio is a major leap forward. It removes the Power Automate middle layer, allowing for faster, smarter, and more intuitive AI experiences. Whether you’re processing orders, connecting to a third-party service, or querying internal APIs, Copilot Studio makes it easier than ever to build powerful, conversational agents.

Happy Developing!