Call ID: 9cc4d61a-aab0-4103-829c-215978ed872b I am designing an AI agent for selling our company's products using vapi. The language I use in the model is Japanese. However, when I call the function to process data with the synchronous option, the AI agent always responds to me with the phrase ちょっと待って (it is an impolite phrase when communicating with customers in Japan). I have tried various ways to set rules so that the AI agent understands and uses more polite expressions, but none have worked. Can you support me? Is there any way to resolve this issue? I look forward to your response. Thank you very much!

@ducnt.ts24 Absolutely, I'd be happy to assist you with this issue. To ensure that your AI agent communicates politely with customers, you'll need to adjust the language model or the response settings to use more appropriate phrases. If you have access to the language model training or configuration, you should look for settings related to language politeness levels or response templates. You can then modify these to include more polite expressions that are suitable for customer interactions in Japanese. If the AI's responses are generated by a specific function in VAPI, you may need to pass additional parameters to that function to control the politeness level, or there might be a separate function designed for handling different politeness levels. Since you've mentioned that you've tried various ways to set rules without success, it might be necessary to review the documentation provided by VAPI for language-specific configurations or to reach out to their support team for guidance on how to implement politeness levels in the AI's responses. If there's a specific function you're calling that's returning the impolite phrase, please provide the name of that function, and I can assist you further with the correct parameters or alternative solutions. In the meantime, if you have any other questions or need further assistance, please let me know!

@User I have tried adjusting the System Prompt in the model multiple times to train the AI to use more formal wording, but it hasn’t worked. It still always says "please wait a moment" before calling my function. This happens not just with a specific function but with any synchronous function I invoke. I want to change this phrase when the AI responds to customers before calling my function, but I don't know where or how to configure it.

@ducnt.ts24 It's an prompting issue, and needs to be written differently. At max here i can try suggesting you different prompts. Can you join today's discord OH?

I'm very interested in this exact topic: @Shubham Bajaj I'd like to better understand the orchestration layer's behaviour when a function is called (sync or async) - does Vapis orchestration layer explicitly prompt the LLM explicitly to tell the user to wait (and thus that prompt needs adjusting, which developers have not got direct access to)?

@ducnt.ts24 because of no response closing this ticket.

@Shubham Bajaj @Ben Thank you for your support. I have resolved this issue. Previously, I used the VAPI UI to create the tools, which caused the AI to use default phrases for requests to the webhook. Afterward, I updated some components using the VAPI API, and it successfully resolved the problem.