How to deploy a Semantic Kernel app with an API endpoint?

[spasegirl]

Hi,

I’m trying to deploy a Semantic Kernel app so that it can be called from other services via an API endpoint.

I went through the official docs, but I couldn’t find a clear guide on how to deploy the app with an HTTP endpoint.

I found this example using Dapr:
https://github.com/microsoft/semantic-kernel/tree/main/python/samples/demos/process_with_dapr
It helps a bit, but I’m still not sure how to set this up fully in Azure.

Could someone please explain or share an example of how to deploy a Semantic Kernel app with an API endpoint?

Thanks a lot!

[moonbox3]

Hi @spasegirl, thanks for the question.

We don’t currently have official docs specifically for deploying SK apps, because SK is a middleware SDK rather than a standalone web framework. It's designed to be embedded into your own applications, so how you deploy it depends on the kind of app you're building.

You can deploy an SK-based app in many different ways, for example, as a microservice behind an HTTP endpoint, an Azure Function, an Azure Web App, a containerized FastAPI app, etc. The Dapr sample you found shows how to kick off an SK process via FastAPI but it isn't related to a deployment.

To move forward, I'd recommend choosing the Azure hosting model that best fits your scenario (for example, Azure App Service for a REST API or Azure Container Apps for microservices) and then following the official Azure documentation for deploying Python apps in that environment. You can then wire up your SK logic inside your preferred web framework (FastAPI, Flask, etc.) and expose the endpoint(s) accordingly.

Hope that helps.