AI Receptionists for Trade Businesses: Building One That's Actually Affordable

The AI receptionist market has a gap in it. Enterprise platforms cost more than a small trade business can justify, and the cheap options are glorified voicemail that cannot book a job or check a calendar.

We built our own system to sit in that gap. This piece explains the problem and the approach, because the reasoning is more useful than the sales pitch.

Why the pricing is broken for trades

Most conversational AI platforms price per minute or per seat, with an enterprise floor attached. That works for a call centre handling thousands of calls a month. It does not work for a two-van plumbing business taking forty calls a week, half of them after hours.

The result is that the businesses with the clearest problem — service trades losing jobs to voicemail every weekend — are the ones the market prices out.

The other half of the problem is integration. A cheap agent that cannot see your calendar or create a job in your system is not solving the actual problem. It is taking a message, and you already had voicemail for that.

What we built

A shared infrastructure layer with per-business configuration on top.

The expensive parts of running a voice agent — the telephony, the speech pipeline, the orchestration, the monitoring — are built once and run at volume. What differs per business is configuration: your greeting, your services, your prices, your escalation rules, your calendar.

That is the whole trick. It is not a technical breakthrough; it is a structural one. Serving fifty trade businesses from shared infrastructure makes the per-business cost a fraction of a dedicated enterprise deployment, without dropping capability.

What each business still gets independently:

• Its own agent brief, voice and greeting
• Its own price list and service definitions
• Its own escalation chain and out-of-hours rules
• Its own calendar and job management integration
• Its own data, separated properly rather than by convention

The integrations that matter for trades

An agent that cannot act is not worth having. For trade businesses the integrations that carry the value are:

Calendar. Real availability, not a booking request someone confirms later. If the agent says Tuesday at 10, Tuesday at 10 must actually be free.

Job management. The job exists in your system before the call ends. Otherwise someone re-enters it, and the time saving evaporates.

SMS. Confirmation to the caller with the callout fee and an ETA. This single thing reduces no-shows and price disputes more than anything else in the stack.

On-call rotation. After hours, the agent needs to know who is actually on tonight, and what to do if that person does not answer.

The emergency question

This is where trade receptionists differ from every other vertical, and where a badly built one is genuinely dangerous.

A burst pipe at 11pm, a gas smell, an electrical fault with water involved — these are not booking requests. The agent's job is to recognise them, give the safety instruction first, and get a human on the phone.

Our rule: safety instruction before booking, always. Turn the isolation valve off. Get out of the house and ring the gas emergency line. Do not touch the switchboard.

An agent that takes a booking for a gas leak has failed regardless of how natural it sounded.

What it costs, honestly

We built this to sit meaningfully below the enterprise platforms while still doing the integrations. What we have not done is make it free, and any vendor offering a capable integrated agent for a token monthly fee is either subsidising you temporarily or not doing the integrations.

The arithmetic that matters is not our price against another vendor's price. It is our price against what you currently lose.

Pull your call logs for a month. Count the misses outside hours. Assume half were real work. Multiply by your average job value. For most trade businesses that number is several times the monthly cost, and if it is not, you should not buy one.

Where we would tell you not to bother

Under about twenty calls a week. The setup effort does not repay at that volume. Use an answering service and revisit in a year.

If your misses are during business hours. That is a staffing problem and an agent addresses the wrong gap.

If you already have leads you are not following up. More capacity at the front door does not fix a leaky middle.

We would rather say this in a discovery call than have you find out in month three.

The comparison, in numbers

Every figure below is an estimate based on typical Australian trade businesses, not a specific client's books. Run it with your own numbers before deciding anything.

Take a two-van plumbing business: roughly 45 calls a week, of which 15 arrive outside business hours.

What a full-time receptionist costs

And they cover roughly 38 hours of the 168 in a week. Every after-hours call still goes to voicemail.

What the agent costs

Covering all 168 hours.

The gap

Roughly $61,000 a year in the first year, and around $64,000 in subsequent years once setup is behind you.

That comparison is deliberately unfair in one direction, so here is the correction: a receptionist does far more than answer calls. They handle the front desk, chase invoices, manage the schedule, and know your regulars. The honest comparison is not agent-versus-person.

The comparison that is actually fair

What are you losing to voicemail right now?

15 after-hours calls a week. Assume half are real work — a deliberately conservative assumption. At an average job value of $640:

• 7 recoverable jobs a week × $640 = $4,480 a week
• Against roughly $180 a week for the agent

Even if only one in five of those calls converts rather than one in two, the agent covers its cost several times over. Below about twenty calls a week, that arithmetic stops working, which is why we tell smaller operators to use an answering service instead.