How do I migrate from a direct Responses API call to an Azure AI Foundry Agent?

Hi,

This is my first software project, and I'm not a professional developer, so please excuse me if I'm misunderstanding how Azure AI Foundry Agents are intended to be used.

I have an Azure AI Foundry Agent that works correctly in Playground.

My application currently uses the Responses API to call a model directly.

The Agent exposes its own Responses Protocol endpoint generated by Azure AI Foundry.

My goal is to have the application use the Agent instead of calling the model directly.

The Agent is already working in Playground and correctly follows the configured system instructions.

My question is:

When an application already uses the Responses API, what is the minimum change required to call a Foundry Agent instead?

• Do I only need to change the endpoint?
• Do I need to create an agent session first?
• Should I use the Agents SDK?
• Or is there another recommended approach?

Does anyone have a JavaScript/TypeScript example of migrating from direct model calls to an Azure AI Foundry Agent?

Thanks!