Full AI Call Center: Handle Customer Service & Sales Calls in One Workflow

This n8n workflow automates both inbound and outbound voice calls, replacing manual call handling that burdens customer service and sales teams. It streamlines tasks like answering customer inquiries or scheduling follow-ups, which often require hours of repetitive phone work. Using Vapi.ai for call handling and OpenAI for AI prompts, it supports dynamic conversations. Key nodes include Webhook (receives Vapi events or outbound requests), Code (detects call type), Code (processes inbound call events like transcripts), Code (initiates outbound calls with AI prompts), and Code (logs results to Slack/Supabase). This helps teams managing 1,000+ monthly calls in sales, support, or booking, enhancing response times and reducing labor costs. Business Value & Requirements Saves 15-20 hours weekly by automating 200+ calls, boosting response rates by 25%. Ideal for mid-sized firms in retail, hospitality, or healthcare. Requires Vapi.ai Pro (~$50/month for 1,000 minutes), OpenAI API (~$0.02/1k tokens), optional Slack/Supabase for logging. Scales to 2,000 calls/month; monitor Vapi rate limits for higher volumes. Needs n8n instance and basic JavaScript knowledge for setup. Technical Setup Install n8n via n8n.io (self-hosted) or cloud.n8n.io (hosted). Get OpenAI API key from openai.com/dashboard/api-keys, Vapi key, phone/assistant IDs from app.vapi.ai/settings. Set environment variables (OPENAI_API_KEY, VAPI_API_KEY, VAPI_PHONE_NUMBER_ID, VAPI_ASSISTANT_ID, optional SLACK_WEBHOOK_URL, SUPABASE_URL, SUPABASE_KEY) in n8n settings. Configure Webhook node with path vapi-call-handler; copy webhook URL from n8n and register in Vapi.ai dashboard. Code nodes use Bearer auth for HTTP requests; ensure valid JSON payloads. Implementation & Maintenance Test with sample inbound webhook from Vapi.ai and outbound JSON (e.g., {context: "schedule demo", goal: "book meeting", name: "Acme", customer_phone: "+12345678901"}). Check logs for transcripts. Common errors: invalid API keys (verify in Vapi/OpenAI dashboards), webhook misconfiguration (test URL in Vapi), rate limits (add delays in Code nodes). Activate workflow in n8n editor; run manually or via webhook. Monitor Vapi usage, rotate keys every 90 days, optimize by caching prompts for high-volume calls. Go live after 10-15 test calls.