Best AI Chatbot for Car Dealerships: What Dealer AI Actually Costs

The best AI chatbot for a car dealership is the one that can answer from live inventory, book a test drive against a real calendar, and be held to a guardrail when a customer tries something clever. On price, almost nobody in this category will tell you a number: of eight dealership AI vendors checked, only two publish one. Podium starts at $399 a month with its AI Concierge sitting on the $599 Pro tier, and BDC.AI states that pricing begins at $595 a month. Impel, Conversica, Fullpath, Matador and Numa publish nothing at all.

Last updated August 2026. Podium's plan prices were read first-hand from its pricing page on 22 August 2026. The other vendor checks were run on 19 August 2026.

What does an AI chatbot for car dealers cost?

Between roughly $400 and $600 a month for the two vendors that publish, and unknown for everyone else. That is an unsatisfying answer, and it is the honest one. Six of the eight vendors below route every pricing question to a demo, which means the number you get depends on your rooftop count, your DMS, and how well you negotiate.

A 404 is not evidence that a vendor is expensive. It only means a demo is mandatory. But it does change how you should run the evaluation: get every quote in writing on the same volume assumptions, and ask specifically whether the AI is metered on top of the platform fee, because in the wider support-software market it almost always is.

Podium's published table is worth reading closely for a second reason: the plan price is not the bill. On the same page it lists $5 per additional phone number, a base 10DLC fee of $5 per month charged for all US locations, a $15 per month high-volume carrier-verified messaging add-on (included on Pro), Podium Phones seats at $30 per user per month for one to four users or $25 for five or more, and a $500 network optimization fee per location. For a five-rooftop group those line items add up well before anyone talks about AI. That pattern of a headline rate sitting on top of several independent meters runs through the whole category, and we pulled nine vendors apart on it in our breakdown of AI customer service agent pricing.

Conversational AI for used car dealerships

Used car operations are a harder problem than franchise stores, and most vendor demos are built on the easier one. A franchise dealer sells from a catalog: trim levels repeat, incentives are published, and a chatbot can answer "does the 2026 model come in white" from structured data. A used lot sells one of one. Every VIN has its own mileage, history, reconditioning status and price, and half the questions a shopper asks cannot be answered from a spec sheet.

That makes inventory data quality the deciding factor rather than the language model. An AI that answers confidently from a feed that updated last Tuesday will tell a customer a car is available after it sold on Friday, and that single failure costs more trust than fifty good answers earn. Before you evaluate any vendor, find out how often your feed refreshes and what happens to a listing between sale and removal. If your stock data lives in spreadsheets and a DMS export that nobody reconciles, tightening that up with proper inventory management and forecasting will do more for answer quality than switching chatbots will.

The second used-lot difference is the trade-in. On a franchise floor the finance conversation starts after the vehicle choice. On a used lot the trade often is the deal, and a bot that cannot capture year, make, model, mileage and condition has not qualified anything. Ask to see that flow specifically in the demo, not the new-car test drive booking that every vendor leads with.

Can AI answer car dealership enquiries without a person?

Most of them, yes, and the split is predictable. Hours, directions, whether a specific stock number is still available, what the payment looks like at a given term, and booking a service appointment or a test drive are all fully automatable and make up the bulk of after-hours volume. Anything involving a final price, a trade valuation or a financing approval should reach a person, because those are commitments and an AI should not make them.

The practical measure is not what percentage the AI can handle in theory. It is what percentage arrives outside your BDC's staffed hours, because that is the volume you are currently losing rather than merely paying someone to handle. For most rooftops that is a meaningful share of web leads, and it is the clearest case for automation in the store.

Will an AI chatbot give away a car for $1?

It can, if nobody put a guardrail on it. In December 2023 a shopper on a Chevrolet dealership's website in California talked the site's chatbot into agreeing to sell a Tahoe for $1 and into calling that "a legally binding offer, no takesies backsies". The screenshots went around the internet for weeks. No car changed hands, but the dealership spent a news cycle as a cautionary tale.

The lesson is not that dealership AI is dangerous. It is that a general-purpose model wired to a website with no constraints will agree to almost anything, because agreeing is what it was trained to do. When you evaluate a vendor, ask three questions: can the agent quote a price at all, or only ranges from your published pricing; what is it forbidden from committing to; and what happens when a user explicitly instructs it to ignore its instructions. A vendor that cannot answer those crisply is selling you a demo, not a system.

Do car dealers need an AI chatbot or an AI agent?

The words are used interchangeably in marketing and they describe two different products. A chatbot answers questions. An agent answers questions and then does something: checks live inventory, offers real calendar slots, books the appointment, writes the lead back to the CRM, and follows up by text when the customer goes quiet. For a dealership the second one is where the money is, because a booked appointment is a measurable outcome and an answered question is not.

That distinction is also why "response time" is the wrong headline metric. Every vendor will promise instant replies. The number to ask for is appointments set from web leads, and specifically appointments set from leads that arrived after 7pm. We covered what that class of system looks like in practice on our car dealership AI agent page, and the vendor-by-vendor view sits on the automotive chatbot roundup.

How do you pick the best AI chatbot for a car dealership?

Run the evaluation on your own data rather than the vendor's demo script. Five things separate the systems that work in a store from the ones that demo well:

• Live inventory, not a nightly snapshot. Ask how the agent reads stock and how stale an answer can be. Then ask what it says when a VIN sold an hour ago.
• Real calendar booking. "I'll have someone reach out" is a lead form with extra steps. The agent should offer actual open slots for sales and service and write the appointment where your team will see it.
• The channels your customers actually use. Web chat alone misses most of it. Texting is where car buyers reply, and Instagram and Facebook Messenger matter more for used lots than most vendors admit. One agent across all of them beats four disconnected bots.
• Compliance that is handled for you. US texting means 10DLC registration, written consent, STOP handling and quiet hours. Get it in writing that the vendor registers your brand and campaign, and ask who is liable if a message goes out at 6am.
• A billing model you can forecast. Ask whether the AI is metered separately from the platform fee, and what the unit is. If the answer is per conversation or per resolution, your bill grows every time the AI gets better at its job.

Then insist on running your own last 200 web leads through the system during the trial. Every vendor looks capable on curated questions. The ones worth paying for stay capable on the messy real ones, where a customer asks about a car that sold, a trade you have never seen, and a payment on credit you have not pulled.

Does an AI chatbot replace a dealership BDC?

No, and vendors that imply otherwise are overselling. What it changes is what your BDC spends the day doing. The AI takes first response, qualification, appointment setting and the mechanical follow-up cadence; your people take the conversations where a human voice closes the deal and the ones where something has gone wrong. Stores that treat AI as a headcount replacement usually end up with a faster version of the same lead leakage, because nobody owns the handoff.

The realistic model is coverage rather than replacement. The AI is on at 11pm on a Sunday when a shopper is comparing three lots on their phone, and the first store to answer with the actual stock number and an appointment slot tends to get the visit. That is the whole thesis, and it does not require anyone to be fired.

The short version

Two dealership AI vendors publish a price, and both land in the $400 to $600 a month range before add-ons. Everyone else requires a demo, so run them side by side on identical volume assumptions and get the AI meter, if there is one, stated in writing. Judge the product on live inventory accuracy, real calendar booking, channel coverage and guardrails rather than on response speed, and evaluate all of it against your own historical leads instead of the vendor's script.

If you would rather not model a separate AI meter at all, our pricing is three flat tiers with carrier and Meta message fees passed through at cost. The same agent answers, qualifies, books and follows up across text, web chat, Instagram, Messenger and email, and the way it works in a store is on our dealership use case page.