A dental clinic in Nagoya ran a quiet experiment last year. For one month, they counted every call that went to voicemail during business hours. The number surprised them: 94 missed calls. They estimated roughly 30 of those were new patient inquiries.
At their average revenue per new patient, those 94 missed calls represented over 600,000 yen in lost potential revenue — in a single month.
They were fully staffed. The problem wasn't people. It was physics: two receptionists can only handle so many simultaneous calls. When all lines are busy, the phone rings, and rings, and nobody answers.
They deployed an AI voice agent three months ago. Today, every call is answered within two rings. The receptionists handle the complex stuff. The AI handles the rest.
What AI Voice Agents Are (And Aren't)
When most people hear "AI on the phone," they imagine the robotic IVR hell of "Press 1 for billing, press 2 for support, press 3 to repeat this menu." That's not what we're talking about.
Modern AI voice agents are conversational. They speak naturally, understand natural language responses, and can carry a real back-and-forth conversation. A caller can say "I'd like to book an appointment for next Tuesday, sometime in the afternoon, for a cleaning" and the AI understands the full sentence, checks the schedule, and confirms the booking — without a single "I didn't understand that, please say yes or no."
They're also context-aware. If someone says "actually, can we do Wednesday instead?", the AI understands it's referring to the appointment just discussed.
Here's what they handle well today:
- Appointment and reservation booking — Checking real-time availability and confirming directly into your calendar or booking system
- FAQ responses — Business hours, location, pricing, what services you offer, parking availability
- Call routing — Understanding why the caller is calling and transferring to the right person
- After-hours coverage — Taking messages and booking appointments when your office is closed
- Outbound reminders — Calling patients or customers to remind them of upcoming appointments, confirm attendance, or follow up on pending actions
The Real Cost of a Missed Call
Most businesses don't measure this. They know they miss calls, but they don't calculate the revenue impact. Here's a rough framework to apply to your own business:
- How many calls do you miss per month? (Check your phone system logs if you're unsure)
- What percentage of those are likely new customer inquiries?
- What's your average revenue per new customer?
For the dental clinic above: 94 missed calls × 30% new patient rate × ~20,000 yen average value = roughly 564,000 yen in monthly opportunity cost.
For a restaurant: 40 missed dinner reservation calls × 60% would-have-booked rate × 5,000 yen average spend × 2 people = 240,000 yen.
The math is different for every business. But almost every business that runs this calculation is surprised by how large the number is.
Where Businesses Are Using This Right Now
AI voice agents are already deployed across a wide range of business types in Japan. The use cases tend to cluster around a few patterns:
Medical and dental clinics High call volume, sensitive scheduling needs, significant revenue per appointment. AI handles new patient booking, appointment confirmations, and basic FAQ. Staff handle insurance questions, medical concerns, and complex rescheduling.
Restaurants and hospitality Dinner reservations are time-sensitive — a caller who doesn't reach you books somewhere else within minutes. AI answers immediately, confirms availability, takes reservation details, and sends a confirmation message.
Real estate Property inquiries come at all hours, especially after people browse listings online in the evening. AI captures lead details, answers basic property questions, and schedules viewings for the next business day.
Beauty salons and spas High repetition in booking requests. Clients often want to book the same service with the same staff member. AI handles the pattern reliably while the stylist focuses on the current client.
Retail and service businesses "Are you open on holidays?" "Do you carry X brand?" "What's the price for Y service?" These questions arrive constantly and are handled identically every time. AI is a natural fit.
What About Japanese Callers?
This is the first question we get when presenting AI voice to Japanese businesses: will it work with natural Japanese speech, including dialects and keigo (formal speech patterns)?
The short answer is yes, with some nuance.
The best current AI voice systems handle standard Japanese very well — clearly spoken sentences, common business requests, and polite Japanese register. They handle regional dialects with decreasing but acceptable accuracy.
What requires more care is genuinely ambiguous speech: elderly callers who expect significant back-and-forth before stating their actual request, highly technical medical questions, or callers who are upset and need emotional acknowledgment first.
For these cases, the AI's most important job is to recognize quickly that the call needs a human, and to make the transfer smoothly. "I want to make sure you get the best help — let me connect you with someone directly" is more important than any specific answer the AI could give.
How a Deployment Actually Works
A typical AI voice agent setup takes 3–4 weeks from kickoff to live calls.
Week 1 — Call audit We analyze your current call patterns: volume by day and hour, most common reasons for calling, current average wait time, missed call rate. This shapes the entire configuration.
Week 2 — Script and persona design We define how the AI introduces itself, what it can and cannot help with, how it handles handoffs, and what happens when a caller goes off-script. The AI has a defined persona — professional, clear, and on-brand for your business.
Week 3 — Integration and testing We connect the AI to your calendar, booking system, or CRM. Internal testing with real call scenarios. Edge case review.
Week 4 — Soft launch and monitoring Live calls with close monitoring. We review call recordings, identify gaps in knowledge or handling, and improve. Most businesses see fewer than 5 calls per 100 requiring any post-deployment adjustments.
Ongoing Monthly review of call logs. Updates when services, pricing, or hours change. The AI learns from patterns over time.
One Thing to Be Honest About
AI voice is impressive. But it's not magic, and the businesses that benefit most are those that go in with clear expectations.
The AI will sound natural. It will handle most calls well. There will be some calls — maybe 10–15% — where a caller is confused, the request is unusual, or the AI isn't confident in the answer. For those, the handoff to a human is more important than the AI's answer.
The metric that matters is not "what percentage of calls does the AI handle alone?" It's "what percentage of callers reach a resolution — either from the AI or from the human it connects them to — without having to hang up and call back?"
When that number goes up, the business wins.
If you're missing calls today, or if your front desk staff spends more than an hour a day on routine call handling, AI voice is worth a close look. We're happy to start with an audit of your current call volume and work out whether the numbers make sense for your business.
