Setting up a custom connector for DeepSeek in Copilot Studio involves several steps that allow you to integrate DeepSeek's advanced AI capabilities into your low-code AI platform. Here's a detailed guide on how to achieve this integration:
Step 1: Understand the Integration Options
Before creating a custom connector, it's essential to understand the available integration options for DeepSeek with Copilot Studio:- Azure AI Foundry: Deploy the DeepSeek R1 model via the model catalog and integrate it using custom topics or conversational boosting with the "create generative answers" action, leveraging Azure OpenAI service integration.
- Custom Connectors: Build a custom connector in Power Platform to integrate DeepSeek or other models within your environment. This allows for secure API calls and controlled access.
- Plugins/REST API: Embed DeepSeek into a Python service or another system running the model. Plugins work similarly to connectors and flows inside Copilot Studio for seamless integration[1][5].
Step 2: Create a Custom Connector in Power Platform
To create a custom connector for DeepSeek, follow these steps:1. Access Power Platform: You can create a custom connector from either Power Apps or Power Automate. Ensure you are using an environment where Dynamics 365 apps are enabled, as environments without Dynamics 365 apps are not supported for custom connectors[4].
2. Set Up Authentication: Use OAuth 2.0 as the authentication type and Microsoft Entra ID (formerly Azure Active Directory) as the authentication provider. This setup secures your back-end service by using two different application registrations[4].
3. Define API Calls: Configure the API calls for your custom connector. This involves specifying the endpoints and methods (e.g., GET, POST) that will interact with the DeepSeek API. Ensure that you handle authentication tokens appropriately to secure these interactions[4].
4. Test the Connector: Once the connector is created, test it to ensure it can successfully communicate with the DeepSeek API. This step is crucial for validating the integration before deploying it in Copilot Studio[4].
Step 3: Integrate the Custom Connector with Copilot Studio
After creating and testing the custom connector, you need to integrate it with Copilot Studio:1. Create an Action in Copilot Studio: Using the custom connector, create an action in Copilot Studio that utilizes the DeepSeek API. This action will enable you to leverage DeepSeek's AI capabilities within your Copilot workflows[4].
2. Enable the Action for Users: Ensure that the action is enabled for your users. This may involve setting permissions or roles within Copilot Studio to control access to the custom action[4].
3. Certify the Connector: Finally, get your custom connector certified. This step is important for ensuring that your integration meets Microsoft's standards and guidelines for security and functionality[4].
Step 4: Deploy and Use the Integration
Once the custom connector is integrated and certified, you can deploy it within your Copilot Studio environment. This allows you to enhance your chatbots with DeepSeek's advanced AI capabilities, such as improved contextual responses and richer user interactions[1][5].By following these steps, you can effectively set up a custom connector for DeepSeek in Copilot Studio, unlocking powerful AI-driven features for your business applications.
Citations:
[1] https://www.linkedin.com/posts/marcel-broschk-85349a1b0_connect-deepseek-in-copilot-studio-activity-7295670990045409281-fAv6
[2] https://blogs.windows.com/windowsdeveloper/2025/03/03/available-today-deepseek-r1-7b-14b-distilled-models-for-copilot-pcs-via-azure-ai-foundry-further-expanding-ai-on-the-edge/
[3] https://www.reddit.com/r/ChatGPTCoding/comments/1e15ig6/make_github_copilot_work_with_deepseekcoder/
[4] https://learn.microsoft.com/en-us/microsoft-sales-copilot/custom-connector-action
[5] https://www.linkedin.com/posts/mirko-fedi-019957174_deepseek-copilot-studio-curious-activity-7290735386480332800-HRcW
[6] https://devblogs.microsoft.com/semantic-kernel/guest-blog-let-your-copilot-declarative-agent-think-deep-with-deepseek-r1/
[7] https://azure.microsoft.com/en-us/blog/deepseek-r1-is-now-available-on-azure-ai-foundry-and-github/
[8] https://www.youtube.com/watch?v=X5654jxAatw
[9] https://dev.to/wassimchegham/deepseek-extension-for-github-copilot-in-vs-code-2c03
[10] https://learn.microsoft.com/en-us/microsoft-copilot-studio/knowledge-graph-connections
[11] https://www.youtube.com/watch?v=YjHmOujfaFs
[12] https://www.youtube.com/watch?v=_kPAreiK0JI
[13] https://www.sharepointeurope.com/events/best-practices-for-connecting-to-external-data-in-copilot-studio/