AI for Auto Repair Shops: Requests and Approvals

By the editorial team airassvet.ru. The guide was prepared on September 29, 2026. Legal requirements were checked against the texts of the Rules for the Provision of Technical Maintenance and Repair Services for Motor Vehicles (Government Decree of the Russian Federation No. 780 dated 05/29/2025, as amended on 10/27/2025) and the Consumer Rights Protection Law in ConsultantPlus as of the same date. The legal sections are for informational purposes only and do not replace legal advice. The editorial team has no commercial relationship with the services mentioned, and there are no affiliate links.

Short answer. AI is most useful in auto repair shops where there is a lot of text and repetitive work: handling requests via messaging apps and by phone, preliminary symptom intake, drafting a repair order and checking it for completeness, photo documentation of the vehicle at intake, and customer-friendly messages about repair status and additional work. Decisions about diagnosis, price, and starting additional work remain with the service advisor and the customer. As of September 1, 2025, the new Rules No. 780 apply: they define what information a service contract must include, when an intake-delivery report can be replaced by photos, and how to approve work that exceeds the estimate — and that is exactly what AI should be configured to support.

Below is what tasks to assign to AI, how to connect them with auto shop documents, where the boundaries are, and how to launch a pilot.

What changed under the new car repair rules
Which auto repair tasks can be handed over to AI
Request intake and repair scheduling
Repair order draft and review
Vehicle intake: photos instead of a report
Additional work and estimates: where AI cannot decide for the customer
Repair status updates and vehicle release
How to launch an AI pilot in an auto repair shop
Pre-launch checklist
Frequently asked questions

What changed under the new car repair rules

By Government Decree of the Russian Federation No. 780 dated May 29, 2025 new Rules for the Provision of Services (Performance of Work) for Maintenance and Repair of Motor Vehicles were approved. The decree took effect on September 1, 2025, and the Rules remain in force until September 1, 2031. The old rules approved by Decree No. 290 of April 11, 2001, have been declared invalid. If your repair order templates and procedures still refer to No. 290, they should be updated — and this is a good time to configure AI around them as well.

For automation, four provisions are important of Section II of the new Rules:

Rule No. 780 clause	What it says	What it means for AI
cl. 6	A request may be submitted in writing or orally, including by phone; the contractor sets the date and time based on the request	AI agent intake via messenger or by voice fits this process
cl. 9	The contract is concluded in writing, and the required list of information is specified	AI can draft and check the repair order for completeness
cl. 10	When a contract is concluded through a website, information system, or app, information may be communicated through them	Messages to the customer through an app or chat are an acceptable communication channel
cl. 12	An intake-delivery report may be omitted if information about completeness and exterior condition, including in the form of digital photos, is included in the contract or an attachment	Photo documentation at intake becomes part of the record

Two more provisions set the boundaries for AI: cl. 18 of the Rules refers to Article 33 of the Consumer Rights Protection Law on estimates, and cl. 14 requires the customer to be notified immediately and work to be suspended in a number of cases, for example if customer-supplied parts are unusable.

Which auto repair tasks can be handed over to AI

The breakdown below is an editorial recommendation based on risk level, not the result of measurements.

Task	AI role	Who makes the decision
Request intake, time selection, booking	Performs	AI within available time slots
Pre-visit symptom intake	Performs	Service advisor during inspection
Repair order draft	Prepares	Service advisor reviews and signs
Repair order completeness check	Performs	Service advisor corrects omissions
Photo documentation at intake	Assists: suggests angles, marks visible damage	Advisor and customer confirm
Explaining the technical diagnosis to the customer	Converts into plain language	The technician approves the wording
Additional work and exceeding the estimate	Prepares the message and records the response	Only the customer gives consent
Diagnosis, repair choice, final price	Does not decide	Technician and customer

Rule for the auto repair shop: AI prepares documents and messages, while signatures, consents, and technical decisions remain with people. That way, AI eliminates routine work without taking on the responsibility that the contractor bears under the Rules.

Request intake and repair scheduling

Most auto repair shop inquiries are repetitive: oil changes, tire service, diagnosis of a knocking sound, routine maintenance. An AI agent in a messaging app or a voice AI on the phone can:

understand a free-form request and clarify the make, model, year, mileage, and symptoms;
check open bays or technician availability in your management system and suggest options;
create an appointment and send a confirmation with the address, time, and list of what to bring;
escalate a complex case to the service advisor along with the data already collected.

Rules No. 780 explicitly allow requests in written and oral form, including by phone (cl. 6). For this, it is important that AI be connected to the real schedule: if it is not, the agent will start promising times that do not exist. The integration logic is the same as for an AI receptionist in the service industry — we covered it using the example of a beauty salon and the booking system API. If requests come in by email, an article on automating email-based requests will be useful automation of requests from email.

Drafting and reviewing the work order

A work order is the written form of the contract. Clause 9 of Rule No. 780 lists what it must include:

contractor information: for a legal entity — company name, registered office and address, OGRN, and registration date; for an individual entrepreneur — full name, registration address, OGRNIP, and registration date;
the client’s full name, phone number, and address;
the contract date and performance deadlines;
the service price, payment method, and payment procedure;
the car’s make, model, main unit numbers, and license plate number;
the list of services, the list of the contractor’s parts and materials, their cost, and quantity;
the client’s parts and materials — with information on mandatory confirmation of compliance, if such requirements are established;
warranty periods, if they are established;
the position and full name of the person who accepted the request, their signature, and the client’s signature.

AI is useful here twice: it builds a draft work order from the conversation and the technician’s notes, then checks the finished document against this list, highlighting missing items. For example: the unit number is missing, the parts list has no quantities, or warranty periods were not filled in even though the shop provides them. This type of checklist is easy to formalize, so it works well for automation — unlike diagnosis.

The draft should not go to the client without a technician’s review. The intake advisor’s and client’s signatures are still required, and it remains a person’s responsibility to ensure the document reflects the actual work and prices.

If your work orders and receiving documents are in 1C, you can connect an agent to it — see the article how to connect AI agents to 1C.

Vehicle intake: photos instead of a form

Under Clause 12 of the Rules, when the client leaves the vehicle, the contractor prepares a handoff/acceptance report with completeness, visible damage, and defects. But the report is not required if the contract states that the client leaves the vehicle, and information about completeness and exterior condition — including in the form of digital photos — as well as about parts supplied by the client is included in the contract or an appendix to it.

This is where computer vision and simple automation help:

the app prompts the intake advisor to take the required angles and will not let them complete intake without them;
the model flags visible damage in the photos, and the intake advisor confirms or corrects the markings;
photos with the date and time are automatically attached to the work order as an appendix.

It is important to understand the limit: a computer vision model may miss a small chip or mistake a glare for damage. That is why AI markings are only a suggestion; the intake advisor and the client confirm the condition. For more on document and image recognition in business processes, see the article OCR and document recognition.

Additional work and estimates: where AI cannot decide for the customer

The most sensitive point in an auto repair shop is when something else is found during the repair. Clause 18 of Rule No. 780 says the estimate becomes part of the contract once the client approves it, and the preparation of the estimate and payment are governed by Article 33 of the Law “On Protection of Consumer Rights”. What matters here:

an estimate can be firm or approximate, and it must be prepared at the request of either the client or the contractor;
the contractor may not increase a firm estimate except in cases of a significant and unforeseen increase in the cost of materials, equipment, or third-party services;
if additional work causes an approximate estimate to be exceeded significantly, the contractor must notify the client in a timely manner; without consent, the client may refuse the contract, and the contractor, if they failed to notify in time, must perform the contract within the approximate estimate.

How to use AI correctly here:

The technician records the finding: what was discovered, why it is needed, how much it costs, and what happens if it is not done.
AI turns the technical note into a customer-friendly message with photos and a new total.
The client replies, and AI records the response with the date and time in the order record.
Work starts only after the client clearly says yes; silence or an ambiguous answer is passed to the technician for a phone call.

A separate case is faults that threaten road safety. Under Clause 21 of the Rules, if the client does not agree to have them fixed, a note about such faults is made in all copies of the handoff/acceptance report, and it is certified by the contractor’s responsible person and the client. AI can remind the intake advisor to make such a note, but it cannot replace it.

Repair status and vehicle release

A common reason for calls is, “So, what’s going on with my car?” An AI agent can answer status questions from your system and send updates on its own: vehicle received, waiting for parts, repair complete, ready for pickup.

When releasing the vehicle, the Rules require the contractor to check the completeness and quality of the work, the condition, and the overall appearance (Clause 24), and the client, together with the contractor, must inspect the vehicle and accept the work (Clause 25). AI can prepare a short checklist for the client on what to review at pickup and remind them that any obvious defects must be noted in the report immediately. The actual inspection and signature remain with people.

Reminders about the next service and service offers are a separate topic: offering new services may be considered advertising, and under Article 18 of the Law “On Advertising” prior consent from the recipient is required. We covered this issue in the article about the AI administrator in the mailing section.

How to launch an AI pilot in an auto repair shop

A process you can repeat in any shop:

Choose one process. For example, handling requests in a messenger app or checking work orders for completeness. Do not launch everything at once.
Record baseline metrics for 2–4 weeks. For requests: how many inquiries there were, how many went unanswered outside business hours, and how many turned into appointments. For work orders: how many documents were sent back for correction and which items were missing.
Build a test set. 50–100 past inquiries or work orders, anonymized, including complex cases: multiple symptoms, client-supplied parts, disputed additional charges.
Test the AI on the set before launch. For each case, note whether the data was gathered correctly, whether all missing items were found, and whether the handoff to the technician worked. Fix the rules and run the set again.
Launch with oversight. During the first few weeks, the service advisor reviews everything prepared by AI. Compare it with the baseline metrics and expand to the next process once the errors are isolated and understood.

The pilot sizes are our recommendation for a small shop; for a multi-location network, the test set should be built for each location separately.

Launch checklist

The work order templates and procedures have been updated to comply with Rules No. 780 instead of No. 290.
AI is connected to the real bay or technician schedule.
The list of mandatory work order details under Clause 9 has been turned into validation rules.
It is defined how intake photos are stored and how they are attached to the contract.
It is specified that additional work begins only after the client gives explicit consent, and the response is recorded with the date and time.
Triggers for escalating the conversation to a technician are defined: a pricing dispute, a complaint, a safety-related fault, or an ambiguous client response.
Service notifications and service offers are separated; consent is required for promotional messages.
Baseline metrics and the test set have been recorded.

Frequently Asked Questions

What can AI do in an auto repair shop?

It can take requests via messaging apps and by phone, schedule appointments for available time slots, ask the customer about symptoms, prepare a draft repair order and check it for completeness, help with photo documentation during vehicle intake, report repair status, and draft clear messages about additional work. Diagnosis, pricing, and approval of the work are determined by people.

Can AI draft a repair order?

AI can prepare a draft from messages and the technician's notes and check the document for the information required by Section 9 of Rules No. 780. The repair order must be signed by the person who accepted the request and by the customer, so the service advisor always reviews the draft.

What car repair rules are in effect now?

As of September 1, 2025, the Rules for the Provision of Maintenance and Repair Services for Motor Vehicles, approved by Resolution No. 780 of the Government of the Russian Federation dated May 29, 2025, are in effect. They replaced the 2001 rules under Resolution No. 290 and apply until September 1, 2031.

Can photos be used instead of a vehicle handoff and acceptance report?

Under Section 12 of Rules No. 780, a report does not have to be prepared if the contract states that the customer leaves the vehicle, and the information about completeness and exterior condition, including in the form of digital photos, as well as the customer's parts, is included in the contract or an attachment to it.

Can AI agree on additional work with the customer?

AI can prepare and send a clear message about the issue found and the new amount, and record the response. Only the customer makes the decision: under Article 33 of the Law on Consumer Rights Protection, the contractor must promptly warn the customer about a significant increase over the estimated price, and without consent the customer has the right to cancel the contract.

How the airassvet.ru team helps implement AI in an auto repair shop

If you want not a separate chatbot, but a connected process from request intake to vehicle handoff, the team at airassvet.ru can:

audit the intake and documentation process: where requests are lost, which repair order fields are most often returned for corrections, and where data is stored;
develop an AI agent for request intake and status notifications integrated with your accounting system or 1C;
implement computer vision for photo documentation during intake with confirmation of the marked items by the service advisor and automated repair order completeness checks.

The first step is to choose one process, such as repair order review, record how it works today, what data and constraints it has, and what criterion will define a successful pilot. If you've chosen a process, discuss the project.

Conclusion

AI for an auto repair shop is not a replacement for the service advisor, but an assistant with paperwork and communication: request intake, drafting and checking repair orders under Section 9 of Rules No. 780, photo documentation during intake, and clear messages about status and additional work. Customer consent for estimates that exceed the budget, signatures, and technical decisions remain with people — that is required by Article 33 of the Law on Consumer Rights Protection and by the Rules themselves. Start by updating your templates to match Rules No. 780, choose one process, record baseline metrics, and test AI on real cases before launch.