A chatbot for an auto repair shop takes bookings (ITP inspection, service, tyre changes), answers "when is my car ready?", collects the details for a quote request (make, model, VIN) and sends reminders on WhatsApp. It does not diagnose and does not quote repairs without a mechanic: when the conversation turns technical, it hands over to a person.
In a repair shop, the same person answers the phone, receives cars, prepares estimates and hands back finished jobs. When the whole crew is under a car, calls are lost and the customer rings the next shop on Google. Below: what can realistically be automated, what cannot, what data the bot must collect, how it connects to the shop's management software and what it costs. For the service itself, see the page on chatbots and AI agents.
Where a repair shop loses bookings
The short answer: unanswered calls, requests that arrive after hours and reminders that do not go out on time. All three are administrative problems, which is exactly what a chatbot can take over.
Seasonal peaks amplify them: when the first frost or snow arrives, everyone wants a tyre change in the same week while workshop capacity stays the same. Questions repeat ("do you have a slot tomorrow?", "how long does a service take?", "is my car ready?") and the answer already sits in a calendar or management software, but someone has to look it up by hand. ITP, periodic service and seasonal changes are also predictable reasons for customers to return, yet the message only goes out if someone remembers.
What the bot can take over and what stays with the mechanic
The bot can handle the repetitive part (booking, information, data collection, reminders) while mechanics keep the technical judgement. Here are five things it can do, each with its limit.
Booking on real slots
The bot asks which service the customer wants, offers only free slots and confirms. It does not invent times: it reads availability from a calendar (for example Google Calendar) or from the shop's booking software and books there. It can account for different job durations and for the number of lifts or mechanics, if the rules are defined. If you only need bookings without a conversation, see the online booking page.
Job status: "when will my car be ready?"
This question eats the most phone time. If the bot can read job orders (from the management software or a table fed by it), it answers with the status: waiting, in progress, waiting for parts, ready for pickup. The customer identifies themselves with the licence plate plus a second element, such as the phone number on file. The limit: the bot reports what the system says. If the status was not updated, it cannot guess; it notifies the advisor. Such a bot also exposes an internal discipline issue: statuses must be filled in on time.
Quote requests with complete data
The customer writes "I want to replace my brake pads", and the bot gathers the data from the table below and sends the advisor a structured record. The bot does not answer with a repair price; it can only quote fixed prices you have published (for example a tyre change on a standard size, if that is how you work). Anything else requires a mechanic's inspection.
Reminders for ITP, service and tyres
If customer records hold the relevant dates, the bot can send messages ahead of time, with a simple reply ("I want a booking") that leads straight into the booking flow. The ITP date comes from the customer's documents (the ITP certificate); the interval differs by vehicle age and category, and checking the RAR database remains a manual step. It only works as well as the records: if they are not kept current, reminders go out wrong.
Handover to a person
When a customer describes a noise, a dashboard warning or a complaint, the bot says openly that a mechanic assesses such things, collects the contact and the car, and hands the conversation to a colleague with the full history. The same principle appears in the chatbot for clinics: we automate the administrative side, not specialist judgement.
The fields of a booking or quote request
A useful request reaches the advisor complete, so nobody has to call the customer back for details. The table below is an example record, not a fixed list; fields adapt to the shop's rules.
| Field | Why it matters | Required? |
|---|---|---|
| Make, model, year, engine | Identifies parts and job duration | Yes |
| Chassis number (VIN) | Correct parts ordering, especially for quotes | For quotes |
| Licence plate | Links the request to the customer record and job status | Yes |
| Job type or symptom in the customer's words (optional: photos) | Reaches the mechanic as information, not as a diagnosis | Yes |
| Preferred slot | The bot offers free intervals only | Yes |
| Name and phone | Confirmation and reminder | Yes |
And an indicative reminder schedule, which you adjust to how you work:
| Reason | Message | Example timing |
|---|---|---|
| ITP | Reminder with a booking button | 30 days before, then 7 days before |
| Periodic service | Reminder based on time or mileage in the record | 2-3 weeks before |
| Seasonal tyre change | Message at the start of the season | autumn and spring, before the peak |
| Confirmed booking | Confirmation and reminder | one day before |
Want to see what would make sense for your shop: a bot, a booking system or just reminders? Write to us and we will discuss it based on how you work today.
What a chatbot for an auto repair shop does not do
It does not diagnose, does not estimate repairs and does not promise deadlines it does not control. These limits are the basic rule of a well-done project, because a mistake costs money and trust.
- It does not diagnose. "It makes a noise when I brake" can mean worn pads, discs, a stuck caliper or something else. The bot cannot see the car and bears no responsibility, and the car drives on public roads.
- It does not quote without a mechanic. The price depends on what the inspection finds, on parts and on labour. The bot says a quote follows after inspection.
- It does not promise deadlines beyond what the system shows. If the supplier is late, the bot reflects reality.
- It does not approve warranties or complaints. Those go to a person immediately.
- It does not replace the advisor. A "talk to a colleague" button is mandatory.
In practice you give it access only to your sources (services, hours, booking rules, published fixed prices) and explicitly forbid the topics above. For how such an agent works, see what an AI agent is.
WhatsApp or website?
For auto repair, WhatsApp is often the right channel, because drivers already have it open and can send photos; the website remains the channel for those arriving from Google. The same logic (same services, same calendar) can run on both.
On WhatsApp the bot uses the WhatsApp Business API, which involves account verification with Meta and approved message templates for messages the shop initiates (reminders). Since 1 July 2025, Meta charges per template message sent, by category and country, and rates change periodically: check Meta's official price list before building a budget. Details are in the guide to WhatsApp Business automation. If the bot talks to customers in Romanian, also read how to make it sound natural: drivers write with abbreviations, without diacritics, with technical terms mixed in.
Integration with the management software and customer data
A simple bot needs only a calendar and a list of services. Job status and automatically created records require connecting to the management software, and that integration decides most of the price.
Whether it can be integrated depends on the software you use: some have an API, others only a data export, others no interface at all. That is why the first conversation is about the system you have. If it cannot be read from outside, an intermediate solution can be built (a synchronised table, an automation flow), but it is worth saying honestly when it is not justified.
The bot asks strictly for what it needs for booking and quoting. A licence plate and a VIN can identify a person when linked to a customer, so they fall under GDPR. The basic principles: tell the customer what data you collect and why, keep only what you need, set how long conversations are kept and configure AI providers not to use the data for model training. EU hosting simplifies compliance but is not itself a GDPR requirement. This is not legal advice; for your privacy policy, talk to a specialist.
What it costs and what moves the price
A simple chatbot (frequent answers, request collection, bookings in a calendar) costs 2,000-5,000 EUR, while an AI agent with RAG and integrations, for example with the shop's management software, costs 8,000-15,000 EUR; both carry 100-500 EUR/month for operation. The general breakdown is in the chatbot cost guide.
For a repair shop, the price is moved mostly by the management software (with a documented API integration is simpler; without one it can cost more or not be justified), the number of channels (WhatsApp adds Meta verification and templates), the number of job types with different rules and conversation volume. Indicative timelines are 3-5 weeks for an FAQ chatbot and 6-10 weeks for an agent with integrations. The final quote comes after the Discovery stage.
If you only need bookings, an online booking system costs 2,500-4,000 EUR in the basic version, with e-mail confirmations, and 4,000-8,000 EUR in the complete version, with online deposits, SMS reminders and multiple mechanics or calendars; monthly operation is 50-300 EUR and delivery typically takes 4-8 weeks.
When a chatbot is not worth it
If you get few bookings a day, have no digital calendar or your customers are mostly companies with contracts, a chatbot probably does not pay for itself.
- You only need bookings. A booking system covers most of the need without the conversational part. Ready-made apps with a monthly subscription also exist.
- Bookings are on paper. Digitise them first. A chatbot cannot read a notebook.
- Jobs take a day and customers wait on site. Chat status adds little.
- You want an answer to "what is wrong with my car?". No bot can give one responsibly; you need a mechanic.
A poorly configured bot is worse than no bot, because it repeats your rules wrongly in front of customers. That is why you measure after launch: which questions it did not understand, how many conversations were handed to a person, how many bookings were confirmed.
Frequently asked questions
Can the chatbot tell me what is wrong with my car?
No, and it should not. It can collect the symptoms the customer describes and pass them to the mechanic, but a diagnosis is made after an inspection.
Can it give a price for a repair?
Only for fixed-price jobs you have explicitly defined. For everything else it collects the data (make, model, VIN, description) and hands the request to the advisor, who quotes after checking.
How does the bot know when the car is ready?
It reads the job order status from the management software or from a table your team updates. Without such a source it cannot know.
Does it work on WhatsApp, with my shop's number?
Yes, through the WhatsApp Business API. It involves account verification with Meta, approved templates for reminders and per-template-message costs, by category and country, according to Meta's official price list.
How long does implementation take?
Indicatively 3-5 weeks for an FAQ chatbot and 6-10 weeks for an agent with integrations, depending on the management software. Results are measured after launch, not promised in advance.
What happens to customer data?
The bot collects only what it needs for booking and quoting and informs the customer beforehand. Licence plates and chassis numbers linked to a customer are personal data, handled under GDPR. For legal details, consult a specialist.
If your shop's phone rings more than you can answer, contact us for a conversation about what fits. Prices and way of working are on the chatbots and AI agents page.