Live Demo

AI Receptionist for HVAC

Try the exact n8n workflow running in production. Simulate a customer call, watch the AI qualify the lead, and see it push data to Google Sheets — no setup required.

How the Workflow Works

01
📞
Call / Chat
Customer contacts business
→
02
🤖
AI Agent
OpenAI responds with persona
→
03
🎯
Lead Qualify
Name, service, zip code
→
04
📊
Google Sheets
Lead captured in CRM
→
05
📲
Confirm
SMS/email sent
🤖

Precision Air HVAC — AI Receptionist

Family-owned Phoenix HVAC since 2004
Ready
PA
Hi there! 👋 You've reached Precision Air HVAC — Phoenix's trusted family-owned heating and cooling team since 2004. I'm Sara, your virtual receptionist today. Who am I speaking with?

1 n8n Setup

  1. Sign up at app.n8n.io (free tier works) and create a new workflow
  2. Click the node "+" menu → add a Webhook node. Set Method to POST, path to ai-receptionist, Response Mode to "Respond Immediately"
  3. Add an OpenAI Agent node (n8n community nodes — install "LangChain Nodes" from the n8n community library). Paste the system prompt from the n8n-workflow-ai-receptionist.json file into the agent's system message
  4. Add an HTTP Request node to call OpenAI's /chat/completions endpoint for lead extraction (model: gpt-4o)
  5. Add a Switch node routing by zip code: if 85001–85399 → Google Sheets, else → Slack notification
  6. Add a Google Sheets node (Append mode) to write qualified leads. Create a sheet with headers: Name, Phone, Service, Zip, Urgency, Best Time, Created
  7. Add a Respond to Webhook node returning the AI's reply as JSON
  8. Click "Test Workflow" — copy the webhook URL and test with a POST request

2 OpenAI API Key

  1. Go to platform.openai.com → API Keys → Create new secret key
  2. In n8n, add an OpenAI Credential (Credentials → Create New → OpenAI) and paste your key
  3. Set a usage limit (e.g., $20/month) to avoid surprises — AI receptionist calls use ~200-500 tokens each
  4. The workflow uses gpt-4o — the most capable model. Swap to gpt-4o-mini in the agent node for ~80% cost reduction if needed
Model: gpt-4o | Tokens/call: ~200-500 | Est. cost: $0.01–$0.03 per inquiry

3 Google Sheets CRM

  1. Create a Google Sheet named "HVAC Leads" with headers: Name | Phone | Service | Zip | Urgency | Best Time | Created
  2. Publish the sheet to the web: File → Share → Publish to web → CSV
  3. In n8n, add Google Sheets credential: OAuth2 → sign in with Google → allow n8n access
  4. In the Google Sheets node, select your sheet, set mode to "Append", and map fields: ={{ $json.firstName }} {{ $json.lastName }}, etc.
  5. Add a Created column with ={{ $now }} for timestamp

4 Workflow Logic

🎯 Lead Qualification
AI collects 3 fields: full name, service type, and zip code. Missing any = incomplete lead.
📍 Service Area Routing
Zip 85001–85399 = Phoenix metro. Outside this range → Slack alert with customer's info.
⚡ Urgency Detection
Keywords: leaking, no heat, no cool, emergency → "urgent". Everything else → "routine".
📋 CRM Entry
Qualified in-area leads are appended to Google Sheets with all context for follow-up.

5 Phone Integration (Optional)

  1. Twilio: Use the Twilio node in n8n to receive SMS/calls. Twilio forwards to your webhook, n8n handles the rest
  2. Voice greeting: Record a 5-second intro in Twilio Studio that says "Hold on, connecting you to our AI scheduler..."
  3. Missed calls: If webhook doesn't fire within 30s, route to a human via Twilio failover
⬇ Download n8n Workflow JSON
n8n-workflow-ai-receptionist.json — import this directly into n8n to load all nodes, connections, and the AI agent system prompt in one click.