M Fazeel
01/13/2026, 7:01 AMM Fazeel
01/13/2026, 7:22 AMChiranjeet Mishra
01/13/2026, 12:09 PMendedReason. If it's not assistant-forwarded-call or shows a pipeline/worker error (e.g., call.in-progress.error-vapifault-* or worker-shutdown), that explains the workflow ending. Docs: https://docs.vapi.ai/calls/call-ended-reason
1. Workflow time limits
• Workflows enforce a 10‑minute engine limit (see related thread). If you’re using a Visual Workflow, it may stop earlier due to internal limits. If you need >10 minutes, run the flow as an Assistant (supports longer maxCallDurationSeconds). See troubleshooting note in threads.
1. SIP/Twilio provider handling
• Vapi may hand off SIP control to Twilio. If Vapi stops executing but Twilio keeps the PSTN leg alive, audio/response will stop. Check the Vapi call object for phoneCallProviderBypassEnabled and phoneCallProviderId. If bypass is true you’re responsible for call control. See transfer/forwarding guide: https://docs.vapi.ai/phone-calling/in-call-control/transfer-calls/debug-forwarding-drops
1. Telephony provider logs
• Use the telephony call ID from the Vapi call and inspect Twilio Call Logs/TwiML execution for disconnects or errors (search CallSid). Docs: https://docs.vapi.ai/phone-calling/in-call-control/transfer-calls/debug-forwarding-drops and Twilio SIP guide: https://docs.vapi.ai/advanced/sip/twilio
1. Worker/keep‑alive or pipeline failures
• If Vapi workers or pipeline fail mid-call (e.g., call.in-progress.error-vapifault-worker-not-available), the workflow can stop while Twilio keeps the leg. Check call logs for pipeline/worker errors and timestamps; correlate with any service outages.Chiranjeet Mishra
01/13/2026, 12:09 PMendedReason, phoneCallProviderBypassEnabled, and phoneCallProviderId.
• Check Twilio CallSid timeline for that CallSid and paste any Twilio error events.