Skip to content

AI in the car repair shop: explaining to customers, approvals, checklists

By the KI können editorial team8 min read

Written with AI and checked against the sources by a second AI. As of November 16, 2026. More in the notes on AI.

In the shop, the writing is rarely why you chose the job: explaining to a customer why the brakes need replacing after all, the message asking for approval, the checklist for the new apprentices. An AI can save you time here. This guide shows where it helps and where you rely only on manufacturer data and your diagnostic tool.

In a car workshop, a mechatronics technician shows a customer something on a tablet next to her car.

Key points

  • An AI explains a finding in words customers understand, if you give it the finding.
  • Tightening torques, fill quantities, settings and test plans come from the manufacturer information, never from an AI chat.
  • For fault codes, an AI can explain in general what a code means. You find the cause with diagnosis and testing on the vehicle.
  • Name, license plate and vehicle identification number don’t belong in a public AI chat, not even on their own.
  • Draft approvals and estimates with placeholders and fill them in in your shop management system.

Which tasks fit in the repair shop

An AI like ChatGPT, Claude, Gemini or Copilot can translate technical language into everyday language, draft messages and sort lists. But it doesn’t see a vehicle, has no access to your diagnostic tool and doesn’t reliably know the manufacturer data.

What you can hand over and what stays with you
Good to hand overStays with you
Explaining a finding clearly for customersThe finding itself
Templates for approval, pickup and appointment remindersPrices, parts and labor times
Explaining in general what a fault code meansFinding the cause on the vehicle
Checklists for the season, vehicle intake and onboardingTightening torques, fill quantities, settings
Drafting replies to reviewsWhether a reply is right and goes online

In the trial lesson you put your first instruction together and get feedback. No account, in a few minutes. Start the trial lesson

Explaining findings clearly

Jonas is a trained car mechatronics technician and works as a service advisor at a car dealership. Many customers don’t understand why a repair is needed and call three times to ask. So he has the AI turn the finding into a short text for the message:

Sample instruction
Explain this finding to a customer with no technical knowledge, formal tone, 80 words at most, no technical terms or with an explanation in brackets.
Finding: front brake discs below minimum thickness, front brake pads still 2 mm. Recommendation: replace front discs and pads.
Say what this means for safety without scaring her and without exaggerating.
Don’t give any prices or times; I will add those myself.
Don’t add any other faults.

Jonas reads the text against the finding. If the AI has added something that wasn’t checked, such as “the brake fluid should also be changed”, he deletes it. A recommendation that doesn’t come from the finding is a sales promise without any basis.

An explanation like this is similar to a conversation at the service desk: you say in simple words what is going on and maybe show the old part.

Placing fault codes without relying on them

Fault codes under the OBD standard, such as P0420, have a general meaning. An AI can explain them and list possible causes. That helps you place a code, especially one you rarely see. It won’t find the cause on the specific vehicle.

Sample instruction
Explain in general what the OBD fault code P0420 means, in three sentences.
Then list possible causes that are typically checked, ordered from common to rare.
Don’t give any readings, target values or tightening torques. Add that testing must follow the manufacturer information.

An AI often doesn’t know manufacturer-specific codes and additional information, or only vaguely. Sometimes it makes up a meaning that sounds plausible. The test plan of your diagnostic tool and the manufacturer’s repair information always come first.

  • Never from the AI: tightening torques, fill quantities, settings, tightening sequences, wiring diagrams.
  • Never from the AI: work steps on high-voltage systems. Only people qualified for it work on high-voltage vehicles, following the manufacturer’s requirements.
  • Never from the AI: approvals for safety-related parts such as brakes, steering or airbags.

Approval, pickup, appointment reminder

Sofia is a master mechanic and runs an independent repair shop with four people. If her team finds something extra during a service, she gets the customer’s approval before the repair. For this, she had templates built that she fills in in the shop management system:

Sample instruction
Write three short message templates from our repair shop to customers, formal tone, 60 words at most each, with placeholders in square brackets.
1. Approval: during the service we found [finding]. According to the estimate, the repair costs [amount]. Please let us know by [time] whether we should go ahead. Without approval, we won’t carry out any extra repairs.
2. Pickup: vehicle [license plate] is ready, pickup from [time] to [time].
3. Reminder: the statutory inspection for [license plate] is due in [month]; book an appointment at [phone number].
Friendly, short, no advertising.

Sofia fills in the placeholders only in the shop management system or her messaging program. Amounts come from the estimate, not from the AI. Whether customers may be contacted by text message, e-mail or messenger, she settles when the job is taken in. (The statutory inspection is the regular roadworthiness test for every car registered in Germany, the Hauptuntersuchung, often just called TÜV after one of the testing organizations.)

Checklists for the season and for onboarding

In fall and spring, almost every shop changes a lot of wheels in a short time. A checklist helps make sure nothing is forgotten even on hectic days. Give the AI your steps and have it sort them and phrase them without gaps:

Sample instruction
Turn my bullet points into a checklist for changing wheels in our shop, to tick off, twelve points at most.
Bullet points: check the job order, get the wheels from storage, check and note tread depth and age, report damage, tighten wheel bolts as specified by the manufacturer, tire pressure as specified by the manufacturer, reset the tire pressure monitoring system if needed, tell the customer to have the bolts re-tightened, label and store the old wheels.
Don’t enter any values. Where a manufacturer specification is needed, write “as specified by the manufacturer”.

For new apprentices, an AI can also turn your bullet points into a checklist for the first week: tools, safety training, contact people. The safety training itself is given by you or your occupational safety specialist, not by a text.

Replying to reviews

Many people look for a repair shop through online reviews and also read the replies. A factual reply to criticism shows how you deal with mistakes. An AI helps you not to sound hurt. Copy the text of the review without the name and add the facts:

Sample instruction
Here is a review of our repair shop: “Appointment at 8 a.m., car not ready until 5 p.m., even though noon was promised. Nobody called.”
What I know: a spare part arrived late, and we forgot to let the customer know. Since this week, we call whenever there is a delay of more than an hour.
Write a public reply, formal tone, 70 words at most. Apologize for the missing call without making excuses.
Mention the new rule. No details about the vehicle, license plate, repair or person, no vouchers.

Before publishing, check that every statement is true. Details about the job that the customer didn’t mention themselves don’t belong in a public reply, even if you know them.

Technical terms for apprentices

Sofia also has the AI explain technical terms for her apprentices, such as what a dual-mass flywheel does. She uses the explanation to start a conversation in the shop, not as study material for the exam. What counts in the exam is in the materials from the vocational school and the motor trade guild (Kfz-Innung).

Customer data and vehicle data

Name, phone number and address are personal data. License plate and vehicle identification number often are too, because they can be linked to a person. Even on their own, they don’t belong in a public AI chat, meaning a freely available service with a personal account and no contract with your business.

  • Instead of “Mr. Kaya, license plate …, 2017 station wagon”, write “a station wagon, built in 2017”.
  • Photos of damage: make license plates and people unrecognizable before you upload a picture, or leave it out altogether.
  • Don’t copy extracts from the shop management system into the chat. Take only the technical details you need.

If you want to work with real customer data, you need a business plan with a data processing agreement (DPA; in German AVV). What that means is explained in the guide ChatGPT for tradespeople.

Check before you send

  1. Does the text match the finding, without extra faults or recommendations?
  2. Does it contain only prices and times that come from the estimate and the schedule?
  3. Is there no technical value anywhere that doesn’t come from the manufacturer information?
  4. Does the text promise anything you can’t deliver, such as “ready today”?
  5. Are name, license plate and vehicle identification number entered only in the shop management system, not in the chat?

With reviews, the same applies as everywhere: an AI may draft replies, but fake or made-up reviews are banned in the EU.

How to practice this

Writing instructions that give the AI exactly the facts it needs, and no more, takes practice. In the app “KI können” you practice this in short lessons, including examples from the trades, and put the instructions to work in your own AI. The trial lesson needs no account.

Sources

Information as of November 16, 2026. Laws and offers can change; this article is not legal advice.