I hosted and presented a session Fronting the Azure AI Stack with Azure API Management at our Nashua Cloud .NET User Group.
We had a great discussion around using Azure API Management as a governed front door for AI services.
The session focused on practical implementation patterns, including how requests flow through APIM, configuring Azure OpenAI / Microsoft Foundry models, authentication and CORS, AI gateway policies, observability, multi-vendor routing, and extending the same gateway pattern to Azure AI Document Intelligence.
We also spent time on the issues that tend to cause the most confusion in real implementations: operation matching, incorrect paths, subscription-key headers, backend authentication, status-code troubleshooting, and keeping provider credentials out of client applications.
The main takeaway was simple: use APIM as a consistent gateway layer so authentication, security policies, usage controls, monitoring, and backend routing can be managed centrally instead of being repeated in every application.
If you couldn’t make it, here’s everything:
Github: https://github.com/nhcloud/ai-gateway
Watch the session: https://youtu.be/uGelt4MIK-M