Will your customers actually talk to an AI on the phone?

Think about the last time you rang a plumber, a dentist, or a mechanic. You didn’t call for the conversation. You called with a task: do you do this, what does it roughly cost, when can you fit me in. The call was a means to an end, and the end was an answer and a time slot.

This is the part the objection misses. Callers aren’t choosing between a lovely human chat and a robot — most of the time they’re choosing between getting their question answered and not getting it answered. A caller who rings at 7pm, asks whether you service their suburb, hears a clear yes, and walks away with a confirmed booking and an SMS in their pocket has had a good phone call. Whether the voice on the other end was a person matters far less to them than whether the call worked.

We already know this from how people behave. Callers happily punch through phone menus, wait on hold, and read from their reference number — not because they enjoy it, but because they want the outcome. What they won’t tolerate is friction without progress: hold music that goes nowhere, a menu that loops, a voicemail that promises a callback that never comes.

Modern AI voices are natural enough that the old objection — “it sounds like a robot from a nineties call centre” — mostly no longer applies. More importantly, a well-set-up AI receptionist isn’t reciting a script; it’s answering from your business’s own information. Asked whether you’re open Saturdays, what a standard service costs, or whether you handle a particular job, it gives your answer, in a normal voice, immediately.

And it can finish the job. The single biggest difference between an AI receptionist and an answering service is that the call can end with an appointment in the diary and a confirmation text on its way — not a message on a pad waiting for you to play phone tag tomorrow.

An AI is only as good as the knowledge you give it. If your prices, hours, and policies aren’t written down anywhere, it can’t answer questions about them — no technology fixes an information problem you haven’t solved yourself.

Some calls need a human. A distressed customer, a complicated complaint, a negotiation, a caller who is elderly or flustered and simply wants to talk to a person — these are not calls to automate, and the right behaviour is a graceful handover, not a stubborn script. Any system you consider should escalate to a human when the call needs one, and take a clean message when it can’t.

And yes, a small number of callers will hang up simply because it’s an AI. That number shrinks every year as people get used to talking to machines everywhere else in their lives, but it isn’t zero, and you should go in knowing that.

The fair comparison is not “AI versus a warm, capable human answering every call” — if you had that, you wouldn’t be reading this. The fair comparison is AI versus what actually happens now: the call that rings out while you’re on the tools, the voicemail nobody leaves a message on, the after-hours caller who dials the next number on their list. Against that baseline, a natural voice that answers instantly, gives real answers, and books the job is not a downgrade. It’s the difference between a captured customer and an invisible one.

If you want to judge it on evidence rather than instinct, that’s the honest way to trial something like Estric: it answers your existing number, books during the call, records and transcribes everything into one dashboard, and hands off to a human when needed — so after a couple of weeks you’re not guessing how callers reacted. You can read every conversation and decide for yourself.