AI in automotive

AI in automotive retail: what runs in production.

Most writing about AI in the automotive industry covers design, factories and the car itself. This page covers the fourth layer: the dealership, where cars are sold and serviced, and where AI agents already carry real customer work. Written from production deployments, not from a forecast.

Updated September 2026

Four layers, one word

"AI in automotive" describes four different things. Engineering AI shortens design and testing with generative design and digital twins. Manufacturing AI runs robots, inspects paint with computer vision and predicts machine failures. In-car AI powers driver assistance and voice assistants. The fourth layer is retail: the showroom, the service desk, the phone line, the WhatsApp thread. It is the layer the customer touches, and the one most articles skip.

Retail AI is different in kind. It does not need a sensor or a factory. It needs the dealer's inventory, the workshop diary and the customer's language, and it has to write real records into systems that already exist.

Where AI is used in the automotive industry

Design and engineering: Generative design, simulation, digital twins. Shortens time to market.
Manufacturing: Assembly robots, visual inspection, predictive maintenance on plant equipment.
In the vehicle: Driver assistance, lane keeping, cabin voice assistants, on-board inference.
Retail and aftersales: AI agents that advise buyers, capture leads, book service into the DMS, follow up, and analyse calls. This page.

What an AI agent does in a dealership

Sales advisory from live stock

A buyer asks for a family SUV under a budget at 11pm, in Arabic. The agent reads the live inventory, answers with real cars, list prices and availability, explains financing options, and captures a test-drive request with a verified phone number. The lead lands in the CRM with the full conversation attached.

Service booking written into the DMS

A customer describes a noise in their own words. The agent maps it to the right service product, reads the live workshop diary for the chosen branch, and writes a real appointment into Keyloop or Autoline with the right advisor and duration. Rescheduling, cancelling and "is my car ready?" run against the repair order.

Follow-up, re-engagement, call analytics

Quiet leads are re-engaged over WhatsApp. Every phone call is transcribed, scored and searchable. When the AI cannot handle something, it hands over to a person with the conversation, the verified details and a suggested next step.

Automotive AI chat: the difference between a chatbot and an agent

Most dealership websites already have a chat window. Most of them are scripted forms with a greeting. The difference is what happens after the customer types.

Website chatbot	AI agent
Scripted decision tree, fixed answers	Answers from live inventory, prices and diaries
Collects a name and hands over a form	Verifies the phone number by OTP and writes the lead into the CRM with context
"An advisor will call you back"	Books the service appointment into the DMS diary, confirmed by SMS
One language, one channel	Arabic, English, German, French; website, WhatsApp, Instagram, phone
Nobody measures it	Every conversation, lead and booking attributed to its source

What it looks like in production

8,000+

customer conversations a month at one Saudi dealer group across BMW, MINI, Ford, Hyundai and a nine-brand marketplace

~70%

of the customer-facing work carried by AI, the rest handed to people with context

~12×

lower cost per conversation than staffing the same coverage

17.4%

word error rate on real BMW service calls in Gulf Arabic; general-purpose engines score 21.5% and worse on the same calls

Ford and Hyundai service branches with one AI front door writing into Keyloop

languages served by a single Swiss dealer's agent: German, French, English

All figures from Salesteq production deployments, see the customer cases. No forecasts.

Who provides AI for automotive retail

The market splits by geography and by what the software writes into. Most named vendors are built around the US dealer stack. Salesteq is built around Keyloop and Autoline networks in Europe and the Middle East, and around languages those systems rarely handle.

Vendor	Home market	Focus
Tekion	United States	Cloud-native dealer management platform with AI across sales, service, finance and operations.
Cox Automotive	United States	AI in marketing, inventory, fixed operations and deal-making across its dealer products.
Salesforce Automotive Cloud	United States	CRM for automotive retailers: lead management, customer profile, sales and service.
Numa	United States	AI for dealership communications and service operations, phone and messaging.
TECOBI	United States	AI-assisted lead response, follow-up and customer texting for dealerships.
Salesteq	Switzerland; live in Saudi Arabia and Switzerland	AI agents that sell and service cars: advisory from live stock, OTP-verified leads, service appointments written into Keyloop and Autoline, WhatsApp and voice, Arabic, English, German and French.

Vendor descriptions are from their public materials, September 2026. Not a ranking.

How to evaluate automotive AI for retail

01Does it write into your DMS diary and CRM, or does it create a parallel list someone has to copy over?

02Does it answer from live stock and list prices, or from a script that goes stale?

03Can it hold the customer's language, including dialect and mid-sentence switches, on voice as well as chat?

04Is the phone number verified before a lead or booking is created?

05What happens when it cannot answer: silent failure, or a handover with the conversation attached?

06Where does the data live, who can see it, and is every AI action auditable?

07Can you measure it: conversations, leads, bookings, cost per conversation, by source?

08How long from contract to the first real booking: weeks, or a transformation programme?

Questions we get on this

What is AI in automotive retail?+

The use of AI agents at the dealership and importer level: advising buyers from live inventory, capturing verified leads, booking service appointments into the dealer management system, following up quiet leads and analysing calls. It sits between the manufacturer's digital tools and the customer.

What is automotive AI chat?+

A conversational agent on a dealer's website, WhatsApp or phone line that answers from live stock and workshop diaries and completes real actions: test-drive requests, financing questions, service bookings, status checks. A chatbot follows a script; an agent reads and writes the dealer's systems.

Does AI replace dealership staff?+

In production it replaces the repetitive traffic: nights, weekends, first answers, bookings. Around 70% of customer-facing work is carried by AI at the group where Salesteq runs; the people handle negotiation, judgment and relationships, with better context than before.

Which dealer management systems can an AI agent write into?+

Salesteq writes real appointments into Keyloop and Autoline workshop diaries today and connects CRM, telephony and messaging (WhatsApp Business, Genesys, Webex, Twilio) through the same integration layer.

Which companies provide AI for automotive retail?+

In the United States: Tekion, Cox Automotive, Salesforce Automotive Cloud, Numa and TECOBI, each built around the US dealer stack. In Europe and the Middle East, Salesteq provides AI agents for dealers, importers and OEM retail networks on Keyloop and Autoline, live at BMW, MINI, Ford and Hyundai retail in Saudi Arabia and at Auto Vogelsang in Switzerland, in Arabic, English, German and French.

How long does it take to go live?+

A pilot on the dealer's own stock and branches goes live in weeks at a fixed price; measurable business value typically lands within 8 to 12 weeks. Each further brand or market takes days once the first is proven.

See it on your own stock.

Your cars, your branches, your languages, live in the system. Not a generic demo.

Book a demo

Or write to [email protected]