When a driver calls an auto repair shop, the first job is not to guess what is wrong with the vehicle. It is to capture enough accurate information for a service advisor to decide what should happen next. A good AI receptionist can answer routine questions, collect vehicle and symptom details, offer a diagnostic appointment, and make sure the shop sees the request. A poor one can promise a price before anyone inspects the car, misclassify a safety concern, or book the wrong kind of visit.

This guide lays out a practical phone intake for independent auto repair shops in the United States. It covers the questions that help a technician prepare, what to say about estimates and repair authorization, how to handle towing and warning lights, and which situations need a human handoff. It is written for shop owners comparing AI answering services, service advisors building a call script, and operators trying to turn more calls into attended appointments.

The Federal Trade Commission advises consumers to request a written estimate and says it should identify the condition to be repaired, parts, anticipated labor, and approval before work exceeds a specified amount. State law may impose additional requirements. That makes an AI receptionist’s boundary clear: collect the concern and book the right next step; do not diagnose, quote a repair as final, or authorize work on behalf of the customer. See the FTC’s Auto Repair Basics guide.

The first minute should reduce friction

A caller usually wants one of four outcomes: a repair appointment, help with a vehicle that may not be safe to drive, an update on a vehicle already at the shop, or a basic answer about hours, location, or services. The receptionist should identify which path applies without forcing every caller through the same long questionnaire. Start with a plain greeting, ask what the caller needs, and offer a person when they request one.

For a new service request, collect the caller’s name, callback number, preferred contact method, vehicle year, make and model, engine or trim only if needed, approximate mileage if useful, and whether the vehicle is at the caller’s location or already at the shop. Ask whether it is drivable, whether it is currently being towed, and what appointment window works. Repeat the phone number and vehicle details before confirming.

Do not assume the caller knows technical terms. “The red oil light came on” is a customer observation; the system should preserve those words rather than translate them into a mechanical diagnosis. Record the exact dash warning, sound, smell, vibration, leak, or performance change in the caller’s language. A service advisor can clarify later. The model year and vehicle identity help the shop prepare; they do not establish the cause or repair.

For existing repair orders, ask for the customer’s name and a safe identifier such as the phone number on file or repair-order number. Confirm the vehicle and ask what update they need. If the system cannot retrieve an accurate status from the shop’s records, it should say that a service advisor will check and call back. It must not invent a completion time or tell the customer that a repair was approved when the record does not show that.

A safe symptom intake without remote diagnosis

Use a short sequence: “What are you noticing?”, “When did it start?”, “Is it happening now?”, “Do you see a warning light or message?”, and “Is the vehicle currently safe and practical for you to drive?” These are information-gathering questions, not a remote inspection. If the caller mentions smoke, fire, a strong fuel odor, brake failure, loss of steering, a collision, or another immediate danger, stop normal intake and follow the shop’s approved emergency handoff. The assistant should not coach a caller to continue driving to the shop.

For a warning light, capture its color, shape, and any text shown on the dashboard, if the caller can describe it safely. Do not ask the caller to photograph a warning while driving. Ask whether the vehicle is stopped in a safe place if the caller says it is on the roadside. If there is a road hazard, collision, injury, fire, or immediate threat, direct them to emergency services or roadside assistance according to the approved script. Do not position the repair shop as an emergency response provider.

For noises or intermittent faults, ask when the issue happens: on startup, while braking, at highway speed, when turning, or only when the engine is warm. Capture the circumstances without proposing a cause. “It squeals when I brake” is useful. “Your brake pads are worn” is an unsupported diagnosis unless a qualified technician has examined the car. Keep the distinction between symptoms and technician findings visible in the repair record.

For a fluid leak, ask where the vehicle is parked and whether the caller can safely identify the fluid or approximate location without touching it. Avoid telling the caller to crawl under a vehicle or handle an unknown fluid. Record whether the leak is active and whether the vehicle is at home, at work, or roadside. The service advisor can determine whether to arrange a tow or schedule an inspection.

Book a diagnostic visit, not a guaranteed repair

The calendar appointment should reflect the work the shop can confidently promise. If the cause has not been confirmed, book a diagnostic inspection or service-advisor consultation rather than a specific repair. A caller might say, “I need a new alternator.” The shop can record the customer’s request, but the AI should not represent that part as the confirmed fix. A diagnostic slot leaves room for inspection, testing, and a written estimate.

Before offering a time, check shop hours, technician capacity, bay or equipment availability, expected appointment duration, and any required pre-visit information. A brake inspection, an emissions test, a tire installation, and an intermittent electrical diagnosis may need different schedules. The system should offer only slots the shop has intentionally made bookable. If it cannot verify the correct appointment type, it should collect preferences and hand the request to a service advisor.

Confirm whether the appointment is a drop-off, a waiting appointment, a tow-in, or a diagnostic-only visit. Ask whether the customer needs transportation options, but do not promise a loaner or shuttle unless the live policy and availability are confirmed. Send the appointment date, arrival instructions, shop address, and cancellation or late-arrival details through the approved channel. The appointment should be saved in the shop’s actual system; a verbal “you are booked” is not enough.

Set expectations honestly: “We can reserve a diagnostic appointment. A technician will inspect the vehicle and the shop will discuss the findings and estimate with you before repairs begin.” This helps avoid the impression that a phone description is a final quote. The FTC’s consumer guidance explains what a written estimate should include and highlights approval before work exceeds the agreed scope. State requirements can differ, so shops should have local counsel or an industry association verify their processes.

Questions the AI must not answer as if it were a mechanic

Do not quote a final repair price from a symptom alone. Parts, labor, vehicle configuration, rust, previous work, diagnostic time, and shop policy can change the estimate. The assistant can read a current published menu price for a clearly bounded service if the shop has approved the wording and conditions. It should label that as a starting or menu price when appropriate, explain what is excluded, and route anything outside the defined service to a person.

Do not promise that a vehicle is safe to drive, that a warning light is harmless, or that a particular repair is required. If the caller describes an urgent safety concern, use a conservative handoff and tell them not to rely on a phone diagnosis. Any immediate roadway danger, injury, smoke, fire, or similar emergency belongs with emergency responders or qualified roadside support, not a conversational agent.

Do not approve work, authorize a charge, waive a diagnostic fee, or change the customer’s repair order unless the system is explicitly connected to a verified authorization workflow and the customer’s consent is captured in the shop’s required manner. A natural-sounding “go ahead” can be ambiguous. The authorization record should identify who approved, what work, the price or limit, and when. If a call is recorded, make sure the shop’s recording disclosure and consent practices meet applicable state law.

Do not make promises about warranties, insurance coverage, recalls, or manufacturer policy based on guesswork. A caller can be directed to the official NHTSA recall lookup, and the service advisor can verify applicability by VIN. Use the official NHTSA recall search rather than relying on a model-year assumption. A recall listing does not mean every vehicle or repair is covered in the same way.

A practical call flow for common repair-shop scenarios

New appointment: identify the vehicle and symptom, confirm whether it is drivable, capture availability, choose the correct diagnostic or service slot, and send a confirmation. The call summary should show what the customer observed and what appointment was booked.

Price question: determine whether the question concerns a published service with a verified menu price or an unknown repair. For a defined service, read the approved price and caveats. For a symptom-based repair, explain that the shop needs to inspect or diagnose the vehicle and route to a service advisor. Never fill a gap with an average found online.

Vehicle currently in the shop: verify the repair order and capture the exact requested update. If the system has no current status, create a callback task with the customer’s preferred contact method and deadline. Do not infer status from the last note if that note may be stale.

Tow or roadside situation: determine whether anyone is in immediate danger and direct emergencies to the proper service. Capture the vehicle location and towing status only if the caller is safe. Transfer to a person if the shop coordinates towing; do not promise a truck, arrival time, or payment arrangement without confirmation.

Complaint or comeback: acknowledge the concern calmly, identify the prior repair and vehicle, and create a priority callback for a manager or service advisor. Do not argue, accept liability, offer a refund, or promise a remedy. Preserve the caller’s words and route according to the shop’s customer-care policy.

Integrating the receptionist with the shop floor

A phone assistant should pass a concise, structured handoff to the person doing the work. Useful fields include caller, callback number, vehicle, mileage if provided, customer-stated symptoms, timing, drivable status, warning-light description, appointment type, scheduled time, towing note, urgency reason, and whether the caller asked for a person. Mark unknown fields as unknown rather than filling them from inference.

Connect the call record to the correct customer and vehicle where the shop management system supports it. If identity is uncertain, create a lead or callback task rather than attaching the conversation to a similar name. Duplicate customer records make repair history unreliable. Test the complete chain from phone call to customer record, appointment, repair order, and follow-up notification.

Define permissions. The assistant may be allowed to create a lead and reserve an appointment but not change labor rates or authorize work. A human should approve exceptions. Keep an audit trail of the original caller statement, any AI-generated summary, and the final staff action. Staff should be able to correct a summary without erasing the source details.

Set a clear outage fallback. If the shop-management connection fails, the assistant should collect a callback request and notify a person, not claim that an appointment has been created. If calendar capacity cannot be checked, do not offer speculative times. Route to the normal voicemail or on-call process and clearly explain what happens next.

How to test an AI answering service before launch

Prepare at least 20 test calls based on actual shop traffic: a routine oil-service request, a warning light, an intermittent noise, an existing repair-order status request, a caller who asks for an exact repair price, a tow-in, a missed call, a caller who changes the vehicle details, an appointment cancellation, a request for a human, and a caller who is upset. Have service advisors score whether the system captured facts accurately and took only authorized actions.

Test interruptions and corrections. Say the vehicle is a 2018 model, then correct it to a 2019. Provide an ambiguous phone number and ask the assistant to repeat it. Give an appointment preference outside business hours. Ask the same question in ordinary language rather than using the script’s exact vocabulary. The agent should recover gracefully and avoid writing the wrong data into the record.

Test the fallback path by disabling an integration in a safe test environment or using a scenario where the agent has no available calendar data. Confirm that a real person is notified and receives the caller’s context. A system should not hide its uncertainty behind confident language. Measure correct appointment creation, field accuracy, human-transfer success, abandoned calls, and customer effort—not just whether the voice sounds natural.

Begin with a small portion of calls, such as after-hours or overflow, and review every record for the first week. Then expand only when the assistant handles the shop’s common scenarios reliably. Keep a staff member responsible for checking failed bookings, call summaries, and callback tasks. Revisit scripts whenever service hours, menu pricing, technician availability, or shop policies change.

What to measure after launch

Track calls answered, calls that reached a human, qualified service leads, appointments booked, appointment show rate, cancellations, calls requiring staff correction, missed callback deadlines, and the share of calls with complete vehicle details. Compare results by call source and time of day. An increased booking count is not automatically improvement if the new appointments are the wrong type or create shop-floor congestion.

Calculate cost per attended appointment and per completed repair, not merely cost per answered call. Include the software fee, phone forwarding or number cost, staff review time, and any incremental service-advisor time. Credit only the appointments that actually show and are appropriate for the shop. Use a conservative estimate of incremental revenue and subtract labor and parts costs before attributing a financial return.

Review a sample of transcripts with service advisors and technicians. Did the summary preserve the symptom in the caller’s language? Did the appointment include the right duration? Was a safety concern escalated? Did the assistant avoid quoting a diagnosis? Track recurring failure patterns and fix the rule or integration behind them. Add a periodic review for seasonal changes and new services.

A service-advisor scorecard for call quality

Score each test call on five dimensions: accuracy of customer and vehicle identity, fidelity to the caller’s own symptom description, correct appointment type, safe handling of urgency and pricing, and completeness of the customer record. Use a simple 0–2 score for each: 0 means failed or unsafe, 1 means partially correct and needing staff repair, and 2 means correct without a correction. Any zero in safety or authorization is a launch blocker even if the average score looks acceptable.

For a new customer, check that the assistant confirmed the callback number and model year, recorded the warning light as described, asked whether the vehicle was drivable, and booked a diagnostic slot rather than a specific repair. For a caller asking for a quote, check that it distinguished a published menu service from an unverified diagnosis. For a comeback complaint, check that the assistant created a manager callback rather than debating fault or offering compensation.

Have two staff members review a subset independently. Differences in their scores reveal ambiguous policies that should be clarified before launch. Keep a short defect log with the call ID, behavior, desired response, owner, and resolution date. Re-run the failed call after a script or integration change to make sure the fix actually worked.

This approach also helps a shop compare vendors fairly. Give each tool the same scenario and scorecard. A polished voice should not earn points for an incorrect booking or unsafe reassurance. The system that creates a reliable service-advisor handoff may be more useful than one that handles a longer conversation but leaves the repair record incomplete.

A repeatable path from first call to approved repair

The first call should establish the customer’s concern and a suitable next step. At the diagnostic visit, a technician inspects the vehicle and reports findings to the service advisor. The shop then prepares the estimate and obtains the customer’s approval through its normal process. If the requested work changes or exceeds the agreed limit, the shop contacts the customer again before proceeding, consistent with applicable law and shop policy.

An AI receptionist can support the first and last administrative steps by booking the diagnostic visit and creating a callback reminder. It should not collapse the inspection, estimate, and authorization into a single phone promise. The FTC recommends a written estimate that identifies the condition, parts, and anticipated labor, and says the estimate should specify approval before work exceeds an agreed amount; state law may require this.

After the repair, the assistant can help with a pickup-hours question or record a request for an invoice copy. It should not tell a customer that the repair is covered by warranty or that a warning light has been fixed unless the current repair record confirms the result. A human should handle disputes, repeat concerns, and requests to change approved work.

Frequently asked questions

Can an AI receptionist diagnose a car problem over the phone?

It should collect the driver’s description and route the caller to the appropriate next step, not provide a definitive mechanical diagnosis. A phone description cannot replace an inspection by a qualified technician.

Can an AI receptionist quote auto repair prices?

It can read a shop-approved, current menu price for a clearly defined service with its conditions. It should not quote a final price for an unknown repair based only on symptoms. The shop should provide a written estimate and capture customer authorization under its process and applicable state rules.

What information should a mechanic call intake capture?

Capture the caller’s contact details, vehicle year/make/model, customer-stated symptoms, when the issue occurs, warning-light description, whether the vehicle is drivable, current location or tow status, and the requested next step. Mark unknown details rather than guessing.

Should an auto shop AI book repairs or diagnostics?

For an unverified issue, book a diagnostic appointment or an advisor-approved inspection type. Reserve a specific repair slot only when the shop has an established rule and enough verified information to do so.

How should an AI receptionist handle a caller who may be unsafe?

Follow a shop-approved safety script, stop routine questioning, and route the caller to emergency services or roadside help where appropriate. Do not tell the caller to continue driving or diagnose the situation remotely.

Sources and further reading

Explore Sysevo’s industry solutions to see how voice workflows connect to a service team’s customer records and next actions.