AI Receptionist for Small Business

Questions people ask

The calls nobody gets to

Most businesses do not have a front desk problem during quiet hours. They have one during the busy ones. The line is engaged, the person who answers is already with a customer, and the caller who needed thirty seconds of information rings the next business on the list instead. Nothing records that this happened, which is what makes it so easy to keep losing.

Add the calls that arrive at lunch, after closing, on a Saturday, and during the fortnight somebody is on leave, and you have a fairly large volume of people who wanted to give you money and could not get an answer. A receptionist agent exists for that volume. It is not there to replace the person at your front desk, who is doing things it cannot do.

What it handles well

• The information calls. Opening hours, parking, whether you take a particular insurance, what a first visit involves, where exactly you are. These are most of the volume and almost none of the value of a trained person's time.
• Qualifying before routing. Establishing what somebody actually needs so the call reaches the right person once, rather than being passed twice and repeated three times.
• Capturing the enquiry. Taking the details of someone who would otherwise have been a missed call with no voicemail, which is the usual outcome.
• Booking. Against your real calendar, covered below.

When it does not know

It says so and hands over. It does not guess, and this is the single most important behaviour in the whole build: a receptionist that invents a plausible answer about your prices, your availability or your policy does more damage in one call than it saves in a hundred.

The handover rules are written with you during discovery, and anything falling outside them escalates by default rather than being attempted. When the handover happens, the conversation so far goes with it, so the caller does not start again from the beginning with a person.

It identifies itself

Callers are told they are speaking to an AI assistant. Some businesses ask whether that hurts conversion. In practice people mind far less about talking to a machine than about being deceived by one, and in several jurisdictions and regulated industries disclosure is not optional anyway.

Booking against a real calendar

Where it books, it reads your live calendar and can only offer slots that genuinely exist. The booking is written by deterministic code rather than produced as text by a language model, which is why it cannot double-book you or offer a time that is not free.

Businesses whose main problem is scheduling rather than answering often want the fuller version of this, with per-service rules, buffers, deposits and no-show follow up. That is the AI appointment scheduler, and the two are frequently built as one agent rather than two.

Voice, web, or both

The same underlying agent can answer the phone, handle web chat, or both, and which you start with should follow where your missed enquiries actually are. For clinics, trades and professional offices that is usually the phone. For ecommerce and software businesses it is almost always the web.

Voice carries a constraint worth knowing before you choose it. A caller notices delay in a way a web visitor does not, so a voice build spends real engineering effort on response latency that a chat build does not need. It is very achievable, but it is not the same project, and anyone quoting you the same timeline for both has not thought about it.

What it will not fix

If callers are leaving because your prices are wrong, your availability is genuinely weeks out, or the experience after the booking is poor, an agent in front of that changes nothing except how quickly people reach the disappointment. We will say so during discovery rather than sell you one and let you discover it at launch.

It also will not handle the call that needs judgement, reassurance or authority. Those are the calls your front desk should be free to take, and clearing the routine volume out of the queue is how they get the time to take them properly.

Related reading

• AI voice agents at the medical front desk
• A voice receptionist for a dental practice in Los Angeles
• AI appointment scheduler: booking, reschedules and no-shows