AI Receptionist Call Grader with Vapi, OpenAI & Google Sheets

Grade every AI receptionist call automatically—before invented info slips through

This n8n workflow acts as your AI Receptionist Call Grader for Vapi calls. When a call ends (or on a weekly schedule), it sends the transcript to OpenAI with your business facts from Google Sheets, scores the call, logs results, and emails (and can text) you when something needs attention.

What this workflow does

• Receives call events via webhook POST when an AI receptionist call ends, or runs weekly on Monday at 7am, or runs manually to create/update the required spreadsheet.
• Parses the end-of-call report to extract transcript, caller, duration, and recording URL—then skips grading for empty or too-short calls.
• Optionally downloads recordings and saves them to a specified Google Drive folder when a recording URL and folder ID are provided.
• Grades with OpenAI by reading the “Business facts” tab from Google Sheets and submitting the transcript + facts for a structured scorecard (outcome, captured details, invented facts, rule breaks, and summary).
• Applies fixed penalties in code to compute a final score/grade, then appends results to the “Calls” sheet and required actions to “Follow-ups”.
• Notifies you fast when a call needs attention: sends a formatted report via Gmail, and can optionally send an SMS alert via Twilio.
• Weekly coaching: summarizes the last 7 days of calls and open follow-ups, asks OpenAI for up to three prompt improvements, and emails the weekly report via Gmail.

Use cases

• Catch invented prices and hours during AI receptionist calls before customers see them.
• Review missed details automatically by enforcing rules from your Google Sheets business facts.
• Operational reporting for SaaS teams: weekly grading + action-oriented follow-ups.

Technical details

• Built with n8n nodes including if, set, code, switch, gmail, and twilio.
• Integrates with Vapi for end-of-call data, OpenAI for scoring, and Google Sheets for “Business facts,” “Calls,” and “Follow-ups.”