"The best way to have a good idea is to have a lot of ideas."

—Linus Pauling, Scientist & Activist

Some dealership content works better when people know what to expect.

A recurring AI character could own one lane, show up consistently, and make service tips, vehicle education, or inventory content feel more like a series.

Before you commit to a face, name, voice, or 37-video content calendar, use this prompt to pressure-test the idea.

AI influencers are getting weird in exactly the way the internet rewards.

Some look like polished fashion models. Others have geometric hair, impossible proportions, villain mustaches, or the general vibe of a Sim who gained consciousness.

And they are pulling real attention. AI characters have built sizable followings, landed brand deals, and generated millions of views. Higgsfield recently made the format easier by letting creators build custom AI characters and place them into video.

A truck-obsessed cowboy can handle towing tips. A mildly chaotic service bot can explain dashboard lights. An EV character can answer charging questions every week without someone having to volunteer for camera duty on Tuesday morning.

That gives stores something traditional dealership content often lacks: a recognizable format people can come back to.

Trying to make an AI influencer look perfectly human can land squarely in uncanny-valley territory. Higgsfield has even leaned toward obviously artificial characters, arguing that flawless faces tend to read as generic AI anyway.

A clearly fictional character gives you permission to exaggerate the personality, build running jokes, and make the AI part of the bit instead of hoping nobody notices.

But, there is one giant asterisk.

Some viral AI accounts are being built by taking real creators' videos, swapping in an AI character, and keeping the performance, gestures, or jokes. Comedians have already called out accounts for copying their material. An AI label does not make borrowed work yours.

Meta dropped two new models this week, Muse Wonder and Muse Wonder Full, and the whole selling point is an assistant that finishes the job instead of answering one question and forgetting why you asked.

Two things make that work. It remembers you between conversations, so the stuff you explained on Tuesday is still there on Friday without a re-explanation. And it sticks with long jobs, trying another angle when the first one stalls and checking in before it does anything you would want to approve first.

Meta also tested it on legal, finance, and health work, which matters more to a store than another coding benchmark does.

One important caveat: all those benchmark wins come from Meta’s own testing, and nobody independent has confirmed them.

Here is why you should care anyway. Meta owns the ad platform your marketing money already lives in and the inboxes most of your social leads come through. Once this model is running Instagram DMs, Messenger, and WhatsApp, the thing chatting with your 11 p.m. shopper is Meta’s, and your BDC vendor is just renting space on top of it. Ask that vendor which model they run and what happens when it changes.

For now, it is in Meta’s developer API and Muse Code, with the consumer app rollout starting with subscribers.

Your dealership website may look great to a shopper and still be nearly invisible to AI.

Wade from Kerzie.ai joins Michael Cirillo to show what ChatGPT is actually looking for when someone asks it to find a vehicle, including the plain-text page that helps AI read your hours, departments, and inventory accurately.

The good news: you can test yours in about 30 seconds.

Watch this one, then check your own site before AI sends a shopper somewhere else.

Thanks for reading, Friend! Remember that even the $300B NFL falls victim to typos from time to time, so don’t be too hard on yourself.