----------------------------- **While I generate an answer, here are some sources that might be helpful**: * [Create Assistant]() * [Orchestration Models]() * [Update Assistant]() * [List Assistants]() -----------------------------

### Answer Hello! I'm sorry to hear about the issues you're encountering with Vapi. Let's address them one by one. 1. **Saving Changes & Slow Loading**: The issue with changes not being saved even after publishing and experiencing slow loading could be related to transient server or network issues. However, it's important to ensure that your internet connection is stable during these operations. If the problem persists, considering the platform's complexity, it might also be linked to temporary system overloads or updates being rolled out. Unfortunately, there isn't direct information from the sources on this specific issue, but maintaining a stable internet connection and trying at different times might mitigate this problem. 2. **Assistant Lagging and Talking Over Itself**: When your assistant talks over itself, it could be due to latency issues or complications with managing turn-taking in real-time conversation flow, which are common challenges in voice AI platforms like Vapi. Vapi implements a custom fusion audio-text model to manage endpointing, which is essential for determining when a user has finished speaking. This mechanism helps in managing turn-taking but can face challenges if the network latency is high or if there is significant background noise [(1)]() . To potentially resolve these issues: - Ensure that the server you are using is geographically close to your target user base to reduce latency. - Check for any background noise that might be interfering with the audio input, as Vapi utilizes advanced noise filtering models to ensure clean audio input. - Monitor the system's usage to ensure it isn't being overloaded, which can cause performance issues and result in the assistant speaking over itself. For both issues, if they persist, consider reaching out to Vapi's support with specific details so t

hey can investigate further, especially if you suspect it might be related to internal bugs or system updates affecting performance. I hope these suggestions help! If you need further assistance, continue monitoring the system and consider specific configurations that might need optimization.

Got it. Tagging @User for additional help.

It is most probably due to the KB. Will provide you update once it is fixed.

and even if you need to modify the code your favourite LLM can explain it and make the changes for you

I do use the Vapi platform, but not via the web application, because I find it unreliable.

my tool deconstructs the assistant JSON into components which can then be edited separately, such as the system prompt, settings, summary prompt, success evaluation prompt, structured data prompt and schema

I save different versions to version control so I can roll back if I later find an issue, and can track changes over time, and when I'm done editing I can use the update command to update a current assistant, or publish to make a new one