Before connecting Curiosity’s AI Agents to an LLM, you must confirm which Copilot configuration your organisation uses, as this determines whether an endpoint is available.
1. Identify Your Copilot Setup
There are only two possibilities:
A) Microsoft-Hosted Copilot
No Azure OpenAI resource appears in your Azure Portal.
No API keys or endpoints are available.
Copilot runs entirely on Microsoft’s internal Azure OpenAI.
You cannot access this endpoint for Curiosity.
In this scenario, you must create your Azure OpenAI resource for Curiosity to use.
B) Customer-Hosted Azure OpenAI
Azure OpenAI resource exists in the Azure Portal.
It contains model deployments, such as gpt-4o.
Endpoint, deployment name, and API key are available internally.
Curiosity can use this endpoint.
To establish your Copilot configuration quickly, speak to:
Microsoft 365 / Copilot Admins – they can confirm the Copilot setup immediately.
Azure Platform / Cloud Engineering Team – they can check whether any Azure OpenAI resources exist.
Network / Security Team – helpful if Azure OpenAI endpoints are private or restricted.
These teams can usually confirm the setup quickly.
2. Accessing the LLM Endpoint
If your organisation is using customer-hosted Azure OpenAI, you’ll need:
Endpoint URL
Deployment name (for example gpt-4o)
API key
Network access from Curiosity to the endpoint
Public: allow outbound HTTPS
Private: update routing / firewall rules
These are all found within the Azure OpenAI resource in the Azure Portal.
3. Opening Access for Curiosity
Once the endpoint is identified:
Confirm that the Curiosity environment can reach it.
Provide endpoint, model name, and API key to the Curiosity admin.
Follow the Curiosity configuration steps linked below.
4. Important Considerations
If Copilot is Microsoft-hosted, the LLM cannot be reused — Microsoft does not expose internal Copilot endpoints.
Creating your own Azure OpenAI resource is the standard path when endpoint reuse is not possible.
Model governance, data residency, and networking policies must align with internal security requirements.
Engage Microsoft 365 admins, Azure platform engineers, and network/security teams early to confirm the correct path.
Once you have access to an OpenAI resource, you can use these instructions on our knowledge base to continue with your configuration.