An AI receptionist that turns missed calls into booked jobs
A complete AI voice product for Australian trades and services — answers on an Australian number 24/7, understands the business, books straight into the calendar, and transfers the calls that need a human.
- Client
- Call Catcher
- Location
- Australia
- Services
- AI & Automation,Product design & development,Cloud & Infrastructure,Cyber Security
Project stage. Shipped and taking real calls. The figures below are product design targets rather than measured results across a customer base — they describe what the platform is built to achieve, and we've labelled them that way rather than presenting them as outcomes already banked.
24/7
coverage including nights, weekends and public holidays
A product capability, not a projection — the agent answers whenever a call comes in.
Under 5 min
from signup to answering live calls
Measured setup time through the onboarding flow, including website scrape and sandbox test.
13
industry templates out of the box
Trades, clinics, salons, hospitality, real estate and more, each with its own language and booking logic.
Overview
Australian small businesses lose leads every day because nobody can get to the phone. Tradies are on a job, clinics are with a patient, salons are mid-appointment, and after five o'clock the calls stop being answered at all.
Call Catcher is an AI receptionist built to fix that — a product Nexcentric designed and built end to end, covering the interface, the cloud infrastructure, the security model and the automation behind it.
Why the obvious solutions don't work
Every business in this position has already tried the alternatives, and each one leaks leads in a different way:
- Phone menus frustrate callers, and a tradie ringing about a quote will simply hang up and call the next number
- Voicemail feels like a dead end — most callers never leave a message, and the ones who do expect a call back that often doesn't come
- Answering services cost real money per call and don't know the business well enough to answer a question
- Generic AI assistants don't understand the trade, the services offered, the pricing, or when a call genuinely needs a human
The brief
Build an AI receptionist that sounds like a real person, answers industry-specific questions, books appointments, and can be set up by a non-technical business owner in minutes.
That last constraint drove most of the design. A product that needs configuring by a consultant is a service business, not a SaaS product — so onboarding had to work for someone standing in a van with a phone.
How it works
We treated this as a product build rather than a chatbot integration: map the caller journey first, choose components against it, then wrap the whole thing in an onboarding flow simple enough that nobody needs to be walked through it.
- 1
The business signs up and the agent learns the business
The onboarding flow scrapes the business website to auto-build a knowledge base — services, pricing, hours, common questions — so the agent starts out knowing what it's talking about rather than needing everything typed in.
- 2
An industry template sets the ground rules
Thirteen templates cover the language, services and booking logic of different trades. A plumber's agent and a hairdresser's agent behave differently out of the box.
- 3
The owner tests it before it answers anything real
An in-app sandbox lets the owner talk to their own agent and hear exactly how it handles questions, before a single customer call reaches it. This is what makes a five-minute setup trustworthy rather than reckless.
- 4
Calls are answered on an Australian number
Twilio subaccounts provision an Australian number per business. Low-latency voice AI through ElevenLabs handles the conversation so it sounds like a person rather than a phone tree.
- 5
Bookings go straight into the calendar
The agent books directly into Google Calendar or Microsoft 365, checking real availability rather than taking a message that someone has to action later.
- 6
Urgent calls reach a human
Configurable rules decide what warrants a warm transfer and to whom. The point isn't to keep humans out of the loop — it's to stop them being interrupted for things the agent can handle.
- 7
Everything is logged and followed up
SMS confirmations, booking reminders and post-call summaries go out automatically, and the owner gets real-time call monitoring, full transcripts and analytics.
What the application does
- 24/7 call answering on a dedicated Australian number
- Natural, low-latency conversational voice
- Industry-specific language, services and FAQ handling
- Direct booking into Google Calendar and Microsoft 365
- Rules-based warm transfer to the right person
- Automated SMS confirmations, reminders and post-call summaries
- Knowledge base auto-built from the business website
- In-app sandbox for testing before going live
- Real-time call monitoring, transcripts and analytics
- Row-level security policies and role-based access
- Australian data residency
- Flat monthly pricing with Stripe billing
What it changes for a small business
The problem was never that owners didn't want to answer the phone — it's that they physically couldn't. The product removes that constraint:
- Calls get answered while the owner is on a job, with a customer, or asleep
- Enquiries become bookings in the calendar rather than voicemails to return
- After-hours and weekend calls stop going to a dead end
- Genuinely urgent calls still reach a person
- The owner can see what was said on every call rather than guessing
- No per-call answering-service bill that scales with success
What it's built on
Included for the technically-minded — and because the stack is part of why it could be delivered quickly and run cheaply.
Frontend
- React
- TypeScript
- Tailwind CSS
- shadcn/ui
Backend & auth
- Supabase
- Row-level security policies
Voice AI
- ElevenLabs
Telephony
- Twilio subaccounts
- Australian numbers
- SMS
Billing
- Stripe
Integrations
- Google Calendar
- Microsoft 365 / Outlook
Automation
- Supabase Edge Functions
- Scheduled jobs
Sandbox first, then let it answer
Handing an AI the front door of someone's business is a significant thing to do. The sandbox exists because an owner should hear exactly how their agent handles a price question or an angry caller before a customer does.
The transfer rules matter for the same reason. The design goal was never to keep humans out of the conversation — it was to stop them being interrupted for the routine calls so they're available for the ones that need judgement.
That's the same line we hold on every AI build: automate the volume, keep the human where the decision matters.
Outcome
What started as a way to stop missed calls became a complete AI receptionist platform — telephony, voice, booking integrations, billing, security and onboarding — built for how Australian small businesses actually operate rather than how software vendors assume they do.
The build covered the full stack: product design, application development, cloud infrastructure, the security model including row-level access policies and Australian data residency, and the automation running the security-critical and scheduled workflows.
Have a process that’s eating someone’s week?
That’s how this one started. Book a free 30-minute review and we’ll tell you honestly whether it’s worth automating — and sometimes the answer is no.