How does WhatsApp automation work for HVAC dispatch?

What actually happens between the customer's message and the technician's phone

A customer sends a WhatsApp message to your business number — often a photo of a leaking unit or a thermostat reading. The automation replies immediately and works through a short, scripted qualification: service address, residential or commercial, equipment type, what the unit is doing, and whether there is no heat or no cooling right now.

Those answers are not left as chat text. They are parsed into fields — address, job type, urgency tier, preferred window — and pushed into your scheduling system through its API. The dispatcher (or the on-call technician directly) receives a job that is already complete enough to route, rather than a voicemail that needs a callback before it can be scheduled.

• Step 1 — Capture: customer messages your WhatsApp Business number, from your website, Google Business Profile or a printed van number.
• Step 2 — Qualify: an AI assistant asks dispatcher-style questions and handles follow-ups, photos and voice notes.
• Step 3 — Route: the structured job is written into your scheduling tool and tagged emergency or routine.
• Step 4 — Confirm: the customer gets an arrival window, an on-my-way message, and a review request after completion.

Why WhatsApp instead of a phone tree or a website chat widget

A phone tree makes an anxious customer wait; a website chat widget only works while they are on your site. WhatsApp keeps the thread open on the customer's own device, so a follow-up two hours later lands in the same conversation with the full history attached — including the photo of the unit.

It also matters geographically. In Karachi and Lahore, WhatsApp is the default channel for contacting a service business at all, so automating it is not an upgrade, it is table stakes. In the USA it is more often a second channel alongside voice and SMS, used heavily for photo-based diagnostics and rescheduling.

What it costs and what drives the cost

There are three separate cost lines and they are worth keeping separate when you compare quotes. First, Meta charges per conversation on the WhatsApp Business Platform, priced by country and conversation category — Pakistan and the USA are billed at different rates, and Meta publishes the current rate card. Second, whatever AI model handles the conversation bills per message. Third, the build itself, which is where quotes diverge most.

Build cost tracks integration depth, not chatbot cleverness. A flow that captures and emails a qualified lead is a small project. A flow that writes jobs into ServiceTitan or Housecall Pro, reads technician availability back out and handles reschedules is a much larger one, because two-way sync means handling conflicts, cancellations and partial failures. Thinkly Solutions scopes and quotes these per business rather than publishing a flat price, because the same feature list costs very differently against a modern API than against a legacy scheduling tool.

What it does not do

It does not diagnose the fault. It collects symptoms and photos so your technician arrives informed, but a refrigerant or electrical diagnosis needs someone on site.

It does not quote a repair price. It can quote published flat-rate items such as a diagnostic fee or a seasonal tune-up, but open-ended repairs should stay with a human.

It does not replace your dispatcher during peak season. It removes the repetitive intake so your dispatcher spends their day on routing decisions and difficult customers.

Who this is not a good fit for

If you run a one-technician shop taking fewer than a handful of calls a week, you will answer those calls yourself faster than any automation pays for itself. If your work is entirely new-construction contracts won by bid rather than inbound service calls, there is no inbound stream to automate. And if your scheduling lives only in a paper book or a shared spreadsheet with no API, the honest first step is moving scheduling into a system that can be integrated — not bolting automation onto something it cannot write to.