AI Agent for Auto Repair Shops: Book Jobs While You Work

Sunday, 08:10, a repair shop in Petah Tikva. Three customers at the service desk, a car on the lift, and the advisor picking up the phone for the fourth time in ten minutes: "how much are new brake pads?". While he checks the model two calls drop, and seven WhatsApp messages from last night still wait. None is a complaint: they all want to book and pay. The person receiving customers and the person answering the phone are the same person, and at peak hour one of them loses.

Why the phone is the bottleneck in a repair shop

Other businesses can say "we will get back to you". Here most inquiries close or break in that first conversation: a customer whose car is stuck is under pressure. Three costs repeat in almost every independent shop:

• Peak hour and inquiry hour are the same hour. Inquiries arrive between 07:30 and 09:30 and after 17:00, exactly when the team is taking cars in and handing them back, and a call missed then rarely comes back. See the five-minute rule on response time and closing odds.
• Interruptions mid-job. "What is happening with my car" is legitimate, but dozens of times a day it stops the advisor and the mechanic.
• No list of who never came back. An unanswered clutch quote, a service pushed to next month: both vanish into the WhatsApp scroll, in a trade where every car returns every 15,000 km and at every annual test.

The six inquiries every repair shop gets

In most independent shops the inbound mix is the same:

1. Booking a routine service. Oil, filters, the 30,000 or 60,000 km service: a calendar question, not a technical one.
2. "How much does it cost". Pads, discs, a clutch, air conditioning: a range by make, model and year, not a figure over the phone.
3. The annual test. What to bring, whether the shop does preparation and how long it takes.
4. Status of a car in the shop. "Ready today? I collect the kids at 16:00".
5. An urgent fault. A car that will not start, a red warning light, a towing request: immediate triage.
6. Approval for work found mid-service. The mechanic opened it up and found the control arm gone.

The first four are information and timing rather than diagnosis, and together they are most of the message volume: the part an AI agent can take end to end, leaving the mechanics and the advisor what needs them.

Three ways to answer when the shop is full

An advisor answering between cars. The most expensive option that never shows in a report: every call stops the intake in front of them, and a long desk queue means an unanswered phone.

An external answering service. Polite, but blind to your price list, bays and calendar: you get a name and a number and still have to call back.

An AI agent on WhatsApp and the website. Answers in seconds, knows the price list, services and models from a live knowledge base, triages urgency, books the slot and opens a customer card. Anything needing judgement goes to the advisor with the history. Deliberately hybrid: the agent takes the volume, the people take the trade.

Booking that understands how busy the shop is

Here a smart agent separates itself from a data-collecting bot. A repair shop calendar depends on free bays, job duration and yard space, not on half-hour windows. Setup defines the rules explicitly: duration by job type (an oil service takes an hour, a clutch two days), number of bays, a daily cap on cars, an intake cut-off and which jobs need the car all day. The agent offers only windows that fit.

The customer then gets a confirmation with the time and what to bring: vehicle licence, second key, and for test preparation what to check first. A reminder goes out the day before, and here it is worth more than elsewhere: a bay left empty by a car that never showed is an hour deleted from the day. See our no-show guide; the scheduling and reminder layer is covered by the AI office manager.

"What about my car": status updates and work approvals

The inquiry that disrupts work most is the easiest to prevent: a customer asks what is happening because nobody told them. Three short proactive messages across the day, "received and in diagnosis", "waiting on a part, expected tomorrow midday", "ready for pickup until 17:00", cut incoming calls without cutting service, and leave the customer better informed than where they have to chase an answer.

The second step is approving extra work. The mechanic reports a finding, the agent sends a short explanation with a price range and asks for explicit approval, and that approval is saved on the customer card with a timestamp. It removes the argument at the desk over "I never approved that" and the wait with a car on the lift. Expensive work, or a customer who prefers to talk, moves to the advisor with the full history. The service layer is covered by the AI support agent, and on Pro you can collect payment inside the chat.

Tests, routine services and the customers who never returned

Most shops sit on an asset they barely use: the customer list from the past two years. A car is a predictable maintenance product, yet most shops wait for the customer to remember.

The agent runs three cycles: the annual test, a month before the licensing date, with an offer to book preparation; the routine service, by date or by the mileage recorded at the previous visit; and quotes that never closed, with a short message after three days and a return on the requested date.

The commercial point: on every plan, follow-ups and workflows are unlimited with no trigger meter, so a shop sending 400 test reminders a month pays what a shop sending 40 pays. It all sits on one customer card in WhaleBiz CRM, so a year later you still see what was done and what was offered but never carried out.

What to prepare before setup

Setup is done by our team, but the agent's quality is decided by your material:

• A price list by job type, with ranges by model and the diagnostic fee.
• A service list and what you do not do, for example bodywork and paint.
• Calendar rules: bays, duration per job type, daily cap on cars and the intake cut-off.
• Opening hours including Fridays and holiday eves, and what counts as an emergency.
• 10-15 recurring questions with your answers, in your words.
• A handover rule: what is urgent, what is complicated, who is a fleet customer.

The setup fee is a one-time payment from 490 ILS depending on scope, takes up to 14 business days from receipt of all materials, and the subscription starts only when the agent goes live. A shop with an existing management system to connect is handled on the Custom plan.

What it costs and when it pays for itself

Solo costs 490 ILS per month: one agent on one channel, 2,000 AI messages (about 300 conversations), a knowledge base of up to ten documents, up to 20 media files and WhaleBiz CRM. Pro costs 990 ILS and adds up to three channels, 5,000 AI messages (about 750 conversations), an extended unlimited knowledge base, up to 100 media files and payment collection in chat. Custom is built to requirement, for example a multi-branch chain. Prices exclude VAT, with the one-time setup fee on top; full details are on the pricing page.

A shop whose inquiries all arrive on WhatsApp can start on Solo; most also get them from the website, Google and Facebook, which makes Pro the natural start. Payback needs two numbers you know: the average invoice and how many inquiries go unanswered each week. At 1,200 ILS an invoice and three missed inquiries a week, that is 12 a month; even if a third would have closed, about 4,800 ILS, several times the subscription. Those numbers illustrate the method, not the market, so put your own in. One last point: Meta charges separately only for business-initiated messages, not for those the customer started, so answering inbound is the cheap part.

Frequently asked questions

Will the agent quote a repair price without seeing the car?

No. It quotes ranges from your price list by job type and by make, model and year, plus the diagnostic fee, and says the binding price follows an inspection. Fixed-price work, such as an oil service for a specific model, can be set as an exact price: a rule written at setup, not a guess.

How does the agent know whether a slot is free today?

The calendar runs on rules you set: number of bays, job duration, how many cars you take in per day and when intake closes. The agent offers only windows free inside those rules, so five services never land in the same hour on two bays.

Can extra work found mid-service be approved in the chat?

Yes, and it is one of the highest-value uses. The mechanic reports the finding, the agent sends a short explanation with a price range and asks for approval, and that approval is stored on the customer card in the CRM with a timestamp. Expensive work goes to the advisor with the history.

Our customers call, they do not type. Is this even relevant?

Some will keep calling, and that is fine. The agents understand and process voice messages sent on WhatsApp, so a customer who dislikes typing can simply record one. Once the shop answers within seconds, part of the customer base moves there and the phone frees up for calls that need a person. The agents are voice-capable too.

What does this cost an auto repair shop?

Solo costs 490 ILS per month: one agent on one channel, 2,000 AI messages (about 300 conversations) and WhaleBiz CRM. Most shops choose Pro at 990 ILS, with up to three channels and 5,000 AI messages. There is also a one-time setup fee from 490 ILS, prices exclude VAT, and the subscription starts only when the agent goes live.