HVAC After-Hours Answering Script: Safety Triage, Call Templates, and On-Call Rules

After the office closes, HVAC phones still ring: no heat on a freezing night, a CO alarm at 2 a.m., a dripping condensate line, or a property manager asking for a morning tune-up. Without a written HVAC after-hours answering script, whoever picks up — a human answering service, an on-call owner, or an AI receptionist for HVAC — improvises. Improvisation is how gas-odor calls get a booking confirmation instead of an evacuate-and-call-911 line.

This page is the standalone script expansion of our HVAC and plumbing setup guide: safety-first triage grounded in public-agency guidance, a full after-hours template, urgency questions, intake fields, fee lines, and a wiring checklist for HighLevel Voice AI or any answering service. Customize the bracketed fields before go-live.

Why does an HVAC company need a written after-hours script?

Because after-hours callers mix life-safety emergencies with routine repairs, and the person answering needs one approved path for each.

Public agencies treat fuel-gas and carbon monoxide concerns as evacuate-first problems, not diagnose-over-the-phone problems. The U.S. Consumer Product Safety Commission (CPSC) says that when a CO alarm sounds, occupants should immediately move outside to fresh air, call emergency services or 911, account for everyone, and not reenter until responders say it is safe (CPSC: Carbon Monoxide Questions and Answers). State and utility gas-safety pages give the same pattern for a gas smell indoors: leave quickly, do not operate lights or appliances, and call 911 from outside (Mass.gov: What to do if you smell gas; PA PUC: Gas Emergencies).

CDC estimates that each year more than 400 Americans die from unintentional carbon monoxide poisoning not linked to fires, more than 100,000 visit an emergency department, and more than 14,000 are hospitalized (CDC: Carbon Monoxide Poisoning Basics). Your script will not prevent every exposure — but it should never keep a CO-alarm or gas-odor caller indoors answering intake questions.

What should the after-hours greeting say?

Open with the company name, disclose that the office is closed (or that this is a virtual assistant), and ask a safety-first branching question before any booking talk.

Greeting template (human or AI):

"Thanks for calling [Company Name]. Our office is closed right now, but I can still help. This is [name / the virtual assistant]. Are you calling about a gas smell, a carbon monoxide alarm, smoke, sparks, or someone who feels ill — or about heating or cooling service?"

Disclosure first, safety branch second, one question at a time. If you use HighLevel Voice AI, put the disclosure in the welcome message and the branching rules in the prompt. HighLevel's setup guide says a reliable prompt should define the agent's role, what to collect, what to avoid, when to transfer, and how to close — and not to guess when information is unavailable (HighLevel: Create, Test, and Deploy a Voice AI Agent).

How should the script handle gas smells and carbon monoxide alarms?

Stop routine intake, give one short approved safety line, tell the caller to call 911 or the gas utility from outside, log what you can safely gather, and alert on-call — do not diagnose.

Gas odor / suspected leak:

"If you smell gas, leave the building right now with everyone inside. Do not turn lights or appliances on or off, and do not use a phone inside. Once you are outside and away from the building, call 911 and your gas utility. I am alerting our on-call team as well. Please call us back from a safe location with your name, callback number, and address when you can."

Align this with your local utility's published instructions. Common themes: evacuate, avoid sparks (no switches, no phones indoors), call from a safe location, do not reenter until cleared (Mass.gov; PA PUC).

CO alarm sounding:

"A carbon monoxide alarm is a serious warning. Get everyone outside to fresh air right now. From outside, call 911. Do not go back inside until emergency responders say it is safe. I am notifying our on-call technician. Please call us back from outside with your name, callback number, and address when you can."

That tracks CPSC's steps: move outside, call 911, head count, no reentry until permission (CPSC CO Q&A). Hard rules: no booking while the caller is still inside; no "check the furnace" or "open a window and wait"; no price quotes; alert on-call with whatever address and callback you already have.

What triage table should on-call and the AI follow?

Use a written four-level table your techs agree on, then paste it into the prompt or answering-service brief word for word.

Your temperature thresholds and fee policy belong in your table. The structure matters more than copying someone else's numbers.

What does a full HVAC after-hours answering script look like?

It is a short branching conversation: greeting → safety check → identity → address → problem → urgency → book or escalate → fee disclosure → recap and text confirmation.

1. Greeting and safety branch. Use the greeting above. If Level 1 triggers, jump to the gas/CO path and stop.

2. Identity. "May I have your full name and the best callback number? I'll read the number back to confirm."

3. Service location. "What is the service address and ZIP code?" If outside your area: "I'm sorry — that address is outside the area we cover after hours. I can [offer referral / take a daytime callback / end politely]."

4. System and symptom. Ask one at a time: heating, cooling, or both; what the system is doing in the caller's words; system type if known (furnace, boiler, heat pump, central AC, mini-split) and approximate age. Collect symptoms. Do not diagnose.

5. Urgency filters. "Is anyone in the home an infant, elderly, or someone with a medical condition that makes temperature risky?" "About what is the indoor temperature, if you know?" "How long has the system been down?" "Is there water leaking from the equipment, or any burning smell that is not a gas odor?" Map answers to your triage table. When unsure between Level 2 and 3 — especially with vulnerable occupants — escalate Level 2.

6. Escalate (Level 2): "I'm treating this as urgent under our after-hours policy. I'm connecting you with / paging our on-call technician now. They will call you at [callback] as soon as they can. Our after-hours service call fee is [amount or 'explained by the tech'], and it [is / is not] credited toward the repair."

6b. Book (Level 3 or 4): "I can schedule the next available [priority / standard] appointment. I have [day/time windows]. Which works better? You'll get a text confirmation at [number]."

7. Recap. "Just to confirm: [name], [callback], [address], [symptom summary], [next step]. Is that correct?"

What information must every after-hours HVAC call capture?

Capture who called, where to go, what failed, how urgent it is, and whether fees were disclosed — every time.

• Full name and confirmed callback number
• Service address and ZIP (plus unit/gate codes if needed)
• Homeowner, renter, or property manager
• Heating, cooling, or both; system type if known
• Symptom in the caller's words
• Triage level (1–4)
• Vulnerable occupants: yes/no
• Indoor temperature if offered
• After-hours fee disclosed: yes/no
• Disposition: transferred, paged, booked (date/time), or message only

In HighLevel Voice AI, use Update Contact Field, Call Transfer, Appointment Booking, Send SMS, and Trigger a Workflow for these steps (HighLevel Voice AI agent guide). For after-hours-only coverage, set Working Hours to evenings/weekends, or use Use as Backup so staff answer first during the day.

What should the script never say?

Never diagnose, invent prices, give DIY safety advice beyond your approved lines, promise arrival times you cannot keep, or pretend to be human.

• No diagnosing. "A tech will assess on site" is the only cause language.
• No unapproved prices. Quote only flat fees listed in the knowledge base. Everything else: "The technician will quote before work starts."
• No improvised safety advice. Gas and CO get the approved lines only.
• No fake ETAs. "They will call you back" beats an unconfirmed "20 minutes."
• No human theater. If it is AI, say so in the greeting.
• No guessing the service area. If the ZIP is unclear, ask again or take a callback.

How do you wire this script into an AI receptionist or answering service?

Paste the triage table and safety lines verbatim, attach a knowledge base with fees and service area, configure transfer and booking actions, then test with real phone calls — including a gas-smell scenario.

1. Agree the triage table with owners and on-call techs.
2. Write approved gas/CO safety lines; have ownership sign off.
3. Load the knowledge base: service ZIPs, hours, after-hours fee, brands, financing wording, what you do not service.
4. Configure actions: Call Transfer to on-call (and a backup), Appointment Booking, SMS confirmation, workflow for a job or opportunity.
5. Set routing: after-hours direct answer, daytime backup, or answering-service handoff with the same script.
6. Add missed call text-back as a safety net for abandoned calls.
7. Test on a real phone. HighLevel notes that Call Transfer does not run in browser Web Call tests, so validate transfers with a Phone Call test (HighLevel docs).

For platform cost context, see the GoHighLevel Voice AI review and the AI Employee Growth vs Unlimited break-even guide. If you are still choosing between AI, a human answering service, and voicemail, use the category cost comparison.

How should you test the after-hours script before go-live?

Run scripted phone tests for every triage level, confirm transfers wake the right person, and read transcripts before the first real overnight shift.

Minimum tests: gas odor and CO alarm (must stop intake); no heat in freezing weather with an elderly occupant (Level 2); no AC in mild weather (Level 3/4); outside service area; unapproved price questions; address changed mid-call; silence/talk-over; spam; transfer to on-call and backup.

Review overnight transcripts daily for the first week, then weekly. Do not change safety lines without ownership approval.

FAQ

Can I use the same script for plumbing after hours? Mostly yes for intake and fees, but swap HVAC symptom questions for plumbing ones (active leak, water near electrical, no water, sewer backup) and keep the same gas/CO evacuate-first rules when fuel appliances are involved. See the HVAC and plumbing AI receptionist guide.

Should the AI diagnose over the phone? No. Collect symptoms in the caller's words and let the technician diagnose on site.

What if the caller refuses to leave during a gas or CO call? Repeat the approved safety line once, urge them to call 911 from outside, and alert on-call with the address you have. Do not coach DIY checks while they stay indoors.

Do I have to disclose an after-hours fee on the phone? If you charge one, say so before the tech is committed. Put the exact wording in the knowledge base.

How long should an after-hours intake call take? Routine booking can finish in a few minutes. Life-safety calls should be short: safety line, escalate, stop.

Is a voicemail box enough after hours? Usually not for Level 1–2 calls. Voicemail cannot give a live evacuate instruction or page a tech. Compare options in our AI vs answering service vs voicemail guide.

Where do I get a fuller prompt with knowledge-base fields? The free Peace Call Prompt Pack includes HVAC prompt and knowledge-base templates you can adapt alongside this script.

Sources (checked October 5, 2026)