← Portfolio  /  Working Build
Working Build · Voice AI Automation

An AI receptionist that books, not just answers

It answers on the first ring, reads your live calendar, books the slot and sends the confirmation, all while the caller is still on the line. At 2am on a Saturday, if that is when the phone rings.

Built and tested end to end. Ready to point at a live business.
The Problem

They do not wait. They call the next business.

Salons, clinics, physios, trades, dental. Four in ten of these calls come in after you have closed.

A missed call is a booking that went somewhere else. In home services, one is worth $300 to $1,200.

Nobody picks up

A Tuesday evening, after closing
9:12 PM
The call comes in. The desk is empty.
9:12 PM
Voicemail. Eight in ten hang up without leaving a message.
9:41 PM
She rings the next place on the list. Three in four callers do.
8:30 AM
You ring back. The slot is already gone.

The agent picks up

Same call, same evening
9:12 PM
Answered on the first ring.
9:13 PM
Two openings offered, read off the live calendar.
9:14 PM
Booked. Confirmation already in her inbox.
8:30 AM
You open the calendar and she is on it.

The difference is not the conversation. It is who has to do something next.

Try It

Take the call yourself

Tap a reply to answer. The right screen shows the n8n branch running, then the calendar and record it writes. Try booking, rescheduling, then cancelling.

Interactive simulation · Brightleaf Dental is a fictional clinic
AV
Ava Brightleaf Dental · front desk
00:00

Call Brightleaf Dental and let Ava handle it.

Every tool the agent calls shows up here, with what it sent and what came back.
Waiting
The n8n branch currently executing
The branch that runs shows up here while it executes.
Brightleaf Dental Aug 24 - 28 · CDT
Brightleaf Dental Ops 0 records
Appointments
Call Recording
Voice Agent
The Shape Of It

Five steps, no human in the loop

Every box happens in the background while the caller is still talking.

1 · Answers

Picks up instantly, any hour, and asks what the caller needs.

2 · Understands

Works out the service and resolves "Thursday morning" into a real date.

3 · Checks

Reads the live calendar, subtracts what is busy, offers two real openings.

4 · Books

Reads the email back letter by letter, then writes the event.

5 · Confirms

Invite to the caller, record to Airtable, transcript logged.

The Machinery

Four things it can do, four branches that do them

The agent holds no logic. Everything it can change sits behind four webhooks.

And branch by branch.

Those four branches do not care where the request came from. The phone is wired to them here; a chat widget or a booking form would hit exactly the same endpoints.

Phone call
Web chat
Online form
The same four webhooks check availability · book · reschedule · cancel
Calendar event
Database record
Confirmation sent
What It Handles

The things people actually ring about

What a caller says, and what the system does about it.

“Can I book something for Thursday?”
Books it. Checks your real calendar, offers only times that are genuinely free, then writes the appointment and sends the confirmation.
“I need to move my appointment.”
Moves it. Finds the existing booking first, so you get one updated appointment rather than two conflicting ones.
“I have to cancel.”
Cancels it. The slot goes back in the diary straight away, and the record stays so you still know it happened.
“What do you charge for that?”
Answers from your settings. If you did not give it a price, it says so instead of inventing one.set up per business
“Can I speak to a person?”
Takes a message. Passes on anything urgent rather than pretending it can help.set up per business

The last two depend on your prices, services and who to escalate to, so they are set up per business. Everything else works the same way whether you are a salon, a clinic or a plumber.

Questions about how this works?

The availability logic, why it will not confirm a booking it cannot verify, or what it takes to point it at your own calendar. My chatbot answers on the spot.

Ask my chatbot →

Missing calls you cannot afford to miss?

I build voice and automation systems that take the repetitive front-desk work off your team. Open to client work and full-time roles.

Get in touch →

AI transparency

On a real call, AI does the talking. Everything after that, checking the calendar and saving the booking, is fixed automation that runs the same way every time.

Figures: small-business call data, after-hours call share, Peerlogic 26-practice study, per-call value by industry. Some are published by companies selling in this space, so treat them as rough.

AI tools helped build this. The design follows an existing approach rather than a new one.