How to Set Up an AI System for Missed Calls and Booking Follow-Up (Like This Sydney Mechanic Did)

Peter Leondis runs three mechanic workshops across Sydney. He built his reputation over 15 years on quality work and customer service. But as the business grew, a specific problem kept getting worse: missed calls.

At one site, a missed call is an inconvenience. Across three, it’s a steady leak of revenue. A customer calls during a busy bay turnaround, no one picks up, they move on to the next workshop in Google Maps. That call cost him nothing to receive and potentially hundreds of dollars to lose.

So Leondis built a fix. He connected an open-source AI model to a voice agent he named “Zoe.” Zoe handles calls outside business hours, answers common questions, and follows up on incomplete bookings with a text. “Even at small volumes,” he told NAB, “it’s already converting leads we would have missed.”

This isn’t a tech story. It’s a revenue story. And the setup is within reach of any Australian small business.

What Zoe actually does

Zoe is a voice AI, which means it answers phone calls using a synthesised voice and a scripted conversation flow. When someone calls Leondis Automotive outside hours, Zoe picks up, greets them, and takes basic information about what they need. If a booking was started but not completed, Zoe sends a follow-up SMS.

The business outcome is twofold: no call goes unanswered, and leads that were cooling off get a prompt nudge. Leondis built it himself using an open-source model. Most business owners won’t do that. They don’t need to.

Tools available to Australian businesses right now

Several services handle this exact use case without requiring technical setup:

• Chaty.ai (Gold Coast-based): AI receptionist built specifically for Australian businesses in tourism and hospitality. Handles calls, answers FAQs, takes bookings. Recently profiled: Chaty.ai: the Gold Coast startup solving tourism’s missed call problem.
• Air AI: US-based but works with Australian numbers. Full conversational voice AI, can handle multi-turn conversations. Pricing starts around US$0.11/minute.
• Smith.ai: Live receptionist plus AI hybrid. More expensive but handles complex calls better. Good for professional services (lawyers, accountants, consultants).
• Shopify cart recovery: Not a phone tool, but worth mentioning for e-commerce owners. The same concept (follow up on incomplete actions) is built natively into Shopify. If you sell online, this is the easiest version of what Leondis built.

For most trades or service businesses, Chaty.ai or Air AI is the right starting point. For professional services, Smith.ai justifies the higher cost.

How to set it up: step by step

This works for any service business that takes bookings or enquiries by phone.

Step 1: Pick a tool. Start with a free trial. Chaty.ai has an Australian-focused onboarding. Air AI has a sandbox mode. Don’t commit to a platform before you’ve heard how it sounds on a real call.

Step 2: Define the hours. Decide when the AI answers vs. when a human does. Most businesses run AI overnight and on weekends. Some route all after-hours calls straight to AI. Start with the hours your team isn’t available and expand from there.

Step 3: Write the scripts. The AI needs to know what to say. Write out three things: a greeting, a list of FAQs with answers, and a call-to-action (book a time, leave your number, visit the website). Keep the FAQ list short. Five to eight questions covers 80% of what most callers ask. You’ll learn the real questions from the call logs.

Step 4: Set up the follow-up trigger. For missed calls specifically, configure the system to send an SMS when a call isn’t answered. Something simple: “Hi, it’s [Business Name]. We missed your call. Can we help? Reply here or call us back on [number].” That SMS alone recovers a meaningful portion of lost leads.

Step 5: Test with real calls. Call your own number. Listen to how the AI sounds. Test edge cases: what happens if someone asks something it doesn’t know? Make sure the fallback is clean. “I’ll pass that to the team and they’ll follow up” is better than a confused silence.

Step 6: Go live and review weekly. For the first month, check call logs every week. Look at what questions came up, what the AI got wrong, and what calls ended without a resolution. Update the FAQ scripts accordingly. The system improves fast when you feed it real data.

What good looks like

A few metrics worth tracking once you’re live:

• Call answer rate. Baseline this before you start. If you’re answering 60% of calls and the AI pushes that to 95%, the gap is your ROI.
• Follow-up conversion. Of the leads that got an automated SMS or follow-up, what percentage booked? Leondis saw conversions from leads that would otherwise have gone cold.
• FAQ accuracy. How often did the AI give a correct answer vs. send the call to a human? Aim for 80%+ on common questions within 30 days.

Who this suits (and who it doesn’t)

This setup works well for businesses that: take a consistent type of enquiry (bookings, quotes, appointment requests), have predictable FAQs, and lose revenue specifically from missed or unanswered calls.

It’s less useful for businesses where every call is different (complex B2B sales, anything requiring negotiation) or where callers expect an immediate human and won’t accept a bot. High-end consulting and legal services often fall into this category.

But for a mechanic, a physio, a plumber, a restaurant, or a tour operator, the fit is strong. The volume is high, the questions are repetitive, and the cost of a missed call is real. That’s exactly the problem Zoe was built to solve.

Sources

• NAB SME AI Adoption Research (2026): AI changing how Australian SMEs get work done
• NAB Research PDF: Benefits from AI (PDF)
• Chaty.ai: SmallBizAI profile

More practical guides: How-To Hub →

📊 Compare AI tools side by side | 💼 Free resources & AI prompt packs

More on AI for hospitality: Hospitality & Food Service Hub →

📊 Compare AI tools side by side | 💼 Free resources & AI prompt packs