Urgent: Calls abruptly ends in workflow by selecti...
# support
k
I’m running a voice workflow assistant that includes a 3-point ID verification node. The node uses API data to verify what the caller says (e.g., DOB, driver’s licence, address). The issue is that the call ends automatically whenever the workflow reaches this 3-ID verification node, even though other nodes work fine. I’ve tested multiple models — GPT-3.1, GPT-4.1, GPT-4-turbo, etc. — but the behaviour is the same. It seems to drop the call right when the verification logic or API matching runs. how to solve this issue so that i can test different model without the issue of call getting ended .
v
Try testing your verification API tool individually in the dashboard to isolate the issue: 1. Go to
Tools > [Your Verification Tool] > Test
in the Vapi dashboard. 2. Send a sample payload that matches what your workflow sends at the 3-point ID verification node. If the tool test fails, check your API endpoint and authentication. If it passes, review your model configuration for the verification node. This step helps determine if the problem is with the tool/API or the model settings ([Debugging voice agents documentation](https://docs.vapi.ai/debugging)). Source: - [Debugging voice agents documentation](https://docs.vapi.ai/debugging)
k
My calls are getting ended with some opena ai models and i am having end reason as : call.in-progress.error-vapifault-openai-400-bad-request-validation-failed
v
I can help you fix the 400 validation error causing your calls to drop during ID verification. I’ll review your VAPI setup, debug the payload and response schema, and adjust how the assistant handles API checks so calls don’t terminate. This will let you safely test any OpenAI model. Could you share the event log or sample payload where the issue occurs? @Kunj
v
I can help you fix the validation error and make your 3-point ID verification node run smoothly without the call ending. The issue likely comes from how the model processes or formats variables during normalization and comparison, causing a malformed request to OpenAI. I’ll review your current prompt logic, restructure the flow to match VAPI’s validation rules, and ensure payloads stay within schema so every model can run safely. Let’s schedule a quick call to go over your setup and debug it together — when’s a good time for you? @Kunj
c
Hi, currently workflows is under development.