Where AI has helped out so far setting up my used car dealership – The AI Car Dealership Project

  • We’ve sold our first car and have been busy buying others from online auctions
  • We’ve got a logo, a website being built and even a corny jingle
  • So far AI has been helping like a toddler does with cooking – but we’ve identified where it could assist more
  • Watch the latest episode of the AI Car Dealership Project above

Time 7:12 am, January 27, 2024

It’s amazing how quickly things have moved with my AI-assisted used car dealership – and I’m quickly beginning to work out exactly what the technology can and can’t help me with.

As you know, I’ve never run a used car business before, so setting up the Clever Car Collection has been somewhat of a baptism of fire.

But because I’m so wet behind the ears it has given me the advantage of looking at things a different way and trying to get AI to help me as often as possible.

It has tried (and failed) to advise me on what bank account to get. I was turned down by seven banks before some bright spark (Jon) told me to try my own personal bank and, hey presto, the problem was solved.

AI has already come up with the business’ name, so I asked ChatGPT to help design the business logo too. It was, well, mildly successful. 

While the concept was pretty solid – a brain above an image of a car – it failed to spell ‘Clever’ correctly, which sort of defeated the object of the brand. 

ChatGPT admitted it wasn’t very good at spelling (I know the feeling) so suggested I enlisted the help of a professional, which I did.

Clever Car Collection logo

In our latest video episode – which you can watch at the top of this post – you’ll see we found another AI-powered website,, which can create ‘songs’. Not that we needed one, but we asked it to write a jingle for the Clever Car Collection nonetheless. 

The result sounded a bit like a synthesised Busted, but the jingle was actually rather catchy. So much so, now whenever I answer the new business batphone it plays in my head. Actually, come to think of it, it might make a good ringtone…

AI has even helped me write the content for my new website. Award winning designers 67 Degrees are helping me produce my first site and I asked AI to produce the words for the About Us page and others. It was a little over the top in places, but with some tinkering was soon ready to fill the site. 

We’ve got a few more tweaks to make before it goes live, but that should be finished soon.

While AI has helped me with my first car sale it wasn’t exactly instrumental in getting it over the line.

I bought a Smart Brabus Fortwo Cabriolet from our friends at specialist dealer EV Experts. They offered me the part exchange and at first I wasn’t so sure it was for me.

At the time I was with Car Quay’s Jamie Caple – the subject of our last video – who showed me the power of Auto Trader’s Retail Rating which said the car scored 89 out of 100. One glance at that, and the prices, and he told me to buy it.

Smart Fortwo Brabus

I’m glad he did. I advertised the car on Auto Trader on Christmas Eve and sold it 14 days later to the second lady who enquired. I worked hard on the pictures and the advert, following what I’d learned from Caple and best practice advice from Auto Trader and it seemed to work.

AI helped me write some quick replies to messages along the way, but its help was mild to say the least.

While AI could have written the original advert for me, I wanted to set a style first. I am a writer, after all, and felt giving AI the task to pen the first one went somewhat against my 25 years in the job. That’s not to say I won’t use it in the future… 


The first sale was a joy. A lovely lady came across one Sunday in early January and took it for a spin. There wasn’t any ‘selling’ involved. I just told her about the car, answered any questions and helped her get used to driving an automatic for the first time. 

We shook hands at the end and she went off to have a think. We did the deal later that evening via WhatsApp and I delivered it to her home two weeks later after it had been through my local workshop for a service and fresh MOT. 

While AI didn’t help too much it did make me realise there were large parts of the transaction it could have dealt with. The initial enquiry stage, serving up the relevant information and finding out what she had to part exchange could have all been carried out by a well-trained AI assistant.

That’s what I really want. An Ironman-style Jarvis assistant that I can coach and eventually trust to speak to customers on my behalf. I want to be able to feedback during or after conversations it has with buyers and tell it what I liked and didn’t like.

I eventually want AI to be able to value part exchanges too – or at the very least gather all the information together by asking the right questions so I can.

It should also be able to deal with all the paperwork. I’ve signed up to use the brilliant DealerKit DMS, which is built by car dealers (and it shows), and I’m sure much of the deal building and emails it sends could be automated by AI and initiated with a few prompts.

Car buying

Another area I want AI to help with is buying cars. As I’m so fresh to this business I have to be led by the data as I have no gut feeling to fall back on. I am currently buying vehicles that Auto Trader’s data suggests would perform the best in my area, after I asked their experts to tell me how to buy more cars like the Smart.

I’ve signed up to a package with the marketplace and have been using its numbers to help me shortlist stock. Following a day of training at their HQ (something they’ll do for any customer, just ask) we’ve used Retail Check, its Market Health metrics and ‘predicted days to sell’ to identify the right cars in my buying profile on sites like Motorway and Carwow.

So far this has been a manual process, but I am sure it is something that can be automated. The dream is for AI to eventually take a feed from every auction house, cross reference it with the pricing and Retail Rating data as well as my own buying preferences to serve me up a daily hit list. I’m working on a plan for that.

I’m quickly realising how stressful buying and selling used cars can be. Anyone who says it’s easy is lying – it’s not.

Judging prep costs from a few pictures in an online auction is hard and some poorly described cars have already led to some difficult conversations with sellers.

In the not-too-distant future I’m going to take the lessons I learned from Caple at the physical auctions and put it into practice myself in the halls.

Currently, I’ve got two cars in stock – a Skoda Citigo and a Nissan Qashqai – both bought because I followed the metrics I’ve outlined above. The Nissan sold yesterday, five days after I advertised it, which is certainly a positive sign.

Next week we’re going to try and buy some more cars and I’m going to see if AI can assist me with the process. Meanwhile, my colleague Jon will be tackling the same task armed with just a laptop and a spreadsheet. We’ll see who gets there first and whether we pick the same cars.

For now, enjoy the latest episode and apologies in advance for the singing…

James Baggott's avatar

James is the founder and editor-in-chief of Car Dealer Magazine, and CEO of parent company Baize Group. James has been a motoring journalist for more than 20 years writing about cars and the car industry.

More stories...

Server 108