how to ask 'required' parameter from tool_calling through forms instead of asking user through plain text.

[Amogh1506]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

model = ChatOpenAI(model="gpt-4o-mini",streaming = True)
model = model.bind_tools(tools=config_tools)

config_tools = [
    {
        "name": "tool_name",
        "description": "<<explaination of the ask>>",
        "parameters": {
            "type": "object",
            "properties": {
                "param1": {
                    "type": "string",
                    "description": "...."
                },
                 "param2": {
                    "type": "number",
                    "description": "...."
                }  
            },
            "required": ["param1","param2"]
        },
     },

Description

I am using LangChain tool calling and binding it to a 4o-mini model. When the user asks questions, the system seems to identify the 'required' parameters from the tool and asks the user for them in plain text. What I want, however, is to ask the user for the inputs through forms based on the format of the 'required' parameters, rather than plain text prompts.

I am trying to trigger an action based on the tool_call identified, but it seems that the AI message doesn't return any tool_call, even though it correctly identifies the list of 'required' parameters. Is there any return value or method that helps me map the LLM response to the specific tool it referred to, so I can trigger form-based input requests?

System Info

.

[eyurtsev]

The .tool_calls attribute will be populated in the AIMessage if the chat model chooses to invoke a tool. It's possible that the model can't figure out whether to invoke a tool or not based on the inputs being provided.

[Amogh1506]

yes, it does not invoke tool. but somehow it asks for the 'required' parameters mentioned in the tool. the point is i want to ask the same 'required' parameters in a customized way.