Custom Gateway support

[jaskaran1994]

My company provides LLM access through a custom Gateway that uses environment variables for model-specific URLs and keys. I want to use this with ChatCompletionAgent and eventually support multi-agent group chats. How do I implement and register a custom ChatCompletionAgent for this setup?

[moonbox3]

Hello @jaskaran1994 are you using Python or .NET? The ChatCompletionAgent requires the use of an AI connector (a model client). You can provide a custom client, depending upon which service you need to use. You can also roll your own AI service depending upon which SK language if you need more customized behavior.