Using semantic kernel with databricks models

I am currently trying to use semantic kernel with databricks hosted sonnet-3.7 model. And I faced with the problem, that when creating an agent with the plugins, the plugins are not executed. It seems like the kernel doesn't support function calling with such models.

I created a class:

class DatabricksSonnetChatCompletion(ChatCompletionClientBase, BaseModel):
    ai_model_id: str
    service_id: str
    endpoint_url: str
    databricks_token: str

    async def complete_chat_async(self, messages, settings=None, **kwargs) -> ChatMessageContent:
        prompt_messages = [{"role": m.role.value, "content": m.content} for m in messages]

        payload = {
            "messages": prompt_messages,
            "temperature": settings.extension_data.get("temperature", 0.2) if settings else 0.2,
            "max_tokens": settings.extension_data.get("max_tokens", 1024) if settings else 1024,
        }

        # ✅ Extract and forward function definitions if provided
        functions = kwargs.get("functions")
        if functions:
            payload["functions"] = functions
            payload["function_call"] = "auto"  # Let model decide

        headers = {
            "Authorization": f"Bearer {self.databricks_token}",
            "Content-Type": "application/json",
        }

        # 🚀 Send the request
        response = requests.post(self.endpoint_url, headers=headers, json=payload)
        response.raise_for_status()
        result = response.json()

        choice = result['choices'][0]
        message = choice["message"]

        # ✅ Handle function call response
        if "function_call" in message:
            func_call = message["function_call"]
            return ChatMessageContent(
                role=AuthorRole.ASSISTANT,
                content=None,
                function_call=FunctionCallContent(
                    name=func_call.get("name"),
                    arguments=func_call.get("arguments")
                )
            )

        # ✅ Normal assistant reply
        return ChatMessageContent(
            role=AuthorRole.ASSISTANT,
            content=message.get("content", "")
        )

but it doesn't work.