Hey everyone, we’re running into a serious issue with the VAPI assistant and integrating our data sources. We’ve tried every possible way to feed it data, but it still doesn’t work properly: • Assistant tab – uploaded files directly there (raw text, PDFs, tables…), but it doesn’t read them correctly. • Query Tool – connected and tested, it shows success and says the query tool was used, but the assistant doesn’t extract the right info. • Vector DB (Trieve) – chunked the data, used an embedding model that supports Serbian, connected it with VAPI — still no accurate results. Details: • Tested with multiple data formats (raw text, PDFs, tables…) • Questions and answers are definitely in the database, but the assistant either skips them or answers incorrectly • It’s for a foreign language (maybe this is causing issues?) Has anyone faced this before or knows what the root cause might be? I can share prompt examples and outputs if needed.

To help track down this issue, could you share: - The call ID - When exactly this happened (the timestamp) - What response you expected to get - What response you actually got instead This would really help us figure out what went wrong!