The Man Who Put On a Peacoat and Went Looking for the Truth About AI Content

He starts with an apology, which is always a good sign that something is about to be taken apart. "What's up, y'all? I'm sorry to burst everybody's bubble," Ross Simmonds tells the room. "Um, but we've all been lied to." Lied to by peers. By gurus on LinkedIn. By colleagues, by CEOs, CMOs, CIOs, CTOs — "you throw in any acronym, we've been lied to by all of them." Even, he adds, by the people we love. Parents. Cousins. Friends. "They all lie."

Watch on YouTube → https://www.youtube.com/watch?v=VXxFJAg7YJw

The examples come fast and personal. The bubblegum that supposedly sits in your stomach for seven years — "literally believed this for way too long." The dome light in the car, which he was convinced until roughly three years ago could get you arrested. He is driving the minivan, he explains, the minivan having replaced a Mini Cooper on account of three kids and an accumulating supply of gray hairs. His daughter flicks the interior light on. He panics. His wife does not. "What are you freaking out for? It's just a dome light." He thought he could go to jail.

And then, having warmed the room with harmless domestic mythology, Simmonds names the lie he actually came to kill: AI-written content will never drive results.

"I see this on LinkedIn every week," he says. "Somebody saying, 'Oh, AI is trash. It's just AI slop. You'll never win if you use AI to write your content. It's garbage. It's garbage.'" It is the received wisdom of an industry currently negotiating with its own obsolescence, and Simmonds — the founder of Foundation Marketing, a Canadian agency that works mostly with B2B software companies — decided he was not going to take it on faith. "I'm a big believer in going out and actually finding answers to questions," he says, "instead of believing all of the headlines and just, like, buying into whatever you read online."

Enter Sherlock Homeboy

The case he picked was the most public failure available. CNET, a publisher producing hundreds of pieces of content daily, had run AI-written articles about money — which is to say, on one of the most competitive and most heavily scrutinized topics on the internet, the kind where Google's double-E-A-T standards for experience, expertise, authority and trust are enforced with unusual vigor. The experiment produced errors. The errors produced headlines. The headlines produced a victory lap.

"Everybody was clapping. Human copywriters were applauding themselves, saying, 'Look at us. Humans forever. We're great. We won. Like, AI does suck. It's trash.'"

Ross Simmonds

CNET announced it was shutting the program down. Case closed. Except Simmonds, in his telling, poured a drink, put on a peacoat, adopted the alter ego of "Sherlock Homeboy" — an aside he pauses to disclaim ("that's not actually me, just in case you're wondering") and then to crowdsource, because he has genuinely been trying to find that plaid coat and reports that AI search has been no help — and went into the traffic tools.

His team went page by page, looking specifically for the citations that disclosed a piece had been written with AI. Then they measured what those pages were still doing. The numbers he put on screen are the load-bearing wall of the entire talk.

"Meaning these pieces of content that they produced using AI were generating organically $1.3 million worth of traffic," he says. "Meaning you could spend it in PPC or you could generate it organically." He gestures at an earlier speaker: "As Mike said, I don't know why we don't invest in this organic thing when it can drive this type of results."

The point is not that CNET's mistakes were fine. The point is that the industry read a story about mistakes and concluded something about capability — and never checked the meter.

A Room Full of People Whose Hands Went Up

Simmonds has been giving talks long enough to have a control group. He remembers standing in front of marketing audiences years ago and asking whether they thought AI could do their jobs. "Marketers were so proud, so happy," he says. Nobody's hand went up. He asks a softer version of the question now — how many of you think AI can do elements of your job — and the hands go up.

His framing of the moment is blunt enough that it lands without a joke attached:

"AI is doing to all of our careers what robotics has long done to blue collar work. Whether you like it or not, that's just the change."

Ross Simmonds

And then, because a keynote cannot survive on dread alone, he runs a quiz. Photos go up on screen. Muffins or Chihuahuas. Bagels or dogs. Fried chicken or puppies. Sloths or pastries. He advances each slide at the speed an AI would have solved it, which is to say instantly, and the room does not keep up. The answers: eight muffins, six bagels, six pieces of chicken, eight pastries. "And guess what?" he says. "AI wouldn't have got hungry. I wouldn't need a break. None of that would have happened."

It's a party trick with a thesis underneath: the machine is already better at the mechanical parts. There is an app now, he notes, that will find Waldo for you. "It's crazy. It's wild. Like, the future is insane."

The Chaos Inventory

Before offering a solution, Simmonds takes inventory of the wreckage, and the list is one every practitioner will recognize. Twitter became X. LinkedIn reach is throttled and depends on the platform's mood; individual humans out-reach the brands that employ them. And Google — the big one — has completed a transition he says he was warning about back in 2018 and 2019.

"You look at Google, it has finally officially become a destination where you get the answers directly from going there," he says. "For years, they told you to create your content and they will send you traffic. Now, that traffic is gone. However, you get a bunch of impressions directly in Google search." He is not sentimental about why. Google, he argues, "cares more about shareholder value than it cares about your domain authority, 35, whatever it is."

Meanwhile the job market has responded to this complexity in the least helpful way available. He puts up a job description demanding PPC, SEO, email, branding, social media, AI tooling, Jasper — six years of experience, forty thousand dollars, and now you have to learn TikTok too, "even though it just rolled out." Budgets are being cut. "Everybody wants Barbie production with Ken budgets," he says. "Go back to 2019. We were chilling. Now we're all stressed out."

He also cannot resist the told-you-so, and he has earned it. He was on stages in 2018 telling marketers to distribute their content to Reddit, YouTube, Quora and LinkedIn. "Everybody in the SEO world said, 'This isn't SEO.'" Those are precisely the domains now showing up in AI citations. "Well," he says, "welcome to the new age."

Four Boxes, Free of Charge

The remedy is a framework he explicitly tells the audience to steal: print it, tape it to your desk, hand it to your team. Four stages, and the industry's historical failure is that it has poured almost all of its energy into the second one.

"If we as an industry can forget the past where we got all caught up in content, content, content — content is king, let's create more content — and put marketing back into content marketing, we will be better off."

Ross Simmonds

On research, his advice is partly about tooling and partly about temperament. Upload the investor report to ChatGPT and ask what it implies for your company. Ask for a competitive landscape returned as a spreadsheet with brand positioning, tone of voice, price, description and audience — an example he draws, endearingly, from his own hunt for electrolyte powders, because "as you can tell, I'm intense. I run around a lot." Say thank you to the model, he adds, "because you don't want the LLM to get you in your sleep one day."

But he insists the tools are not the edge. "I think that's one of the competitive advantages for good marketers," he says, "is that we actually have taste."

The Lavender Problem

To demonstrate taste, he invents a client: Rossy's Gardening Blog. He does not have a green thumb and says so. He goes to YouTube rather than a keyword tool and finds a video called Easy Basil Propagation with 350,000 views. That's a signal. He works outward — propagating rosemary, propagating lavender — and the keyword tool tells him lavender does somewhere between 500 and 850 searches a month. Decent. Not thrilling.

Then he goes back to YouTube and adds up what people are actually watching on propagating lavender from cuttings. The videos total roughly 455,000 views.

"That's massive," he says. "That means I should probably consider creating some content for this, because 455,000 views is way more substantial than what the keyword told me about 800 visits on a monthly basis."

The Money You're Spending in the Wrong Place

Here the talk turns from clever to uncomfortable, because Simmonds starts naming line items. His agency works primarily with B2B software companies, and those companies pour thousands of dollars a year into review sites. "It's wild," he says. "Why is it wild? Because when you look at the traffic, tons of sites are getting tanked."

G2: tanking. Capterra: tanking. TrustRadius: tanking. Softwareadvice.com: tanking. Some sites are thriving — he singles out a calculator site as an exception — but the category's direction of travel is down, and the traffic is going to places like Reddit, which is being cited across the large language models.

"You invest thousands of dollars into these review sites. However, you sleep completely on sites like Reddit. You sleep completely on sites like YouTube. It is absolutely mind-blowing to me."

Ross Simmonds

He says his team has been monitoring top bottom-of-funnel SaaS queries across Capterra, G2 and the rest — alongside Reddit — since 2024. "Reddit has officially taken over all of them. There's no question." And in Google's AI Overviews, using data he attributes to the team at Profound, the recurring citations are YouTube, and increasingly LinkedIn. His suggested response is not complicated: maybe invest in LinkedIn articles. Maybe empower your executives to publish articles and newsletters there. Meanwhile the review sites account for under two percent of citations by his figures — "1.5%" — "however, we continue to funnel hundreds of thousands of dollars into these businesses regularly."

His credibility here is unusual, and he offers it as a joke that is also a résumé: "I've been banned from Reddit about 18 times. I love it. It's great. We have a love-hate relationship."

The Coogi Sweater, and Other Machine Errors

On creation, Simmonds is candid about what the tools do well and what they do badly — including to his own face. Small team? Upload your dormant YouTube back catalog to ChatGPT and ask it to generate thumbnails. You'll get something decent. In his case you also get a rendering of him wearing a Coogi sweater while discussing lavender propagation, which he flags himself: "I agree." He asks the model to switch it up. The result still doesn't quite look like him.

"Listen," he says, "this is the worst AI will ever be."

The full pipeline he demonstrates is worth listing because it is entirely assembled from off-the-shelf parts: ask ChatGPT for a script on propagating lavender; take that script to HeyGen; connect ElevenLabs to clone the voice. He notes his clone struggles with the Canadian "aboots," and reports it sounds about ninety percent like him — an assertion the room has to take on faith, because the audio fails to play live. ("Do we have any audio? ... Find me later.") From there, Descript can push the same video into Mandarin, Spanish, French, German or Italian in seconds. "So you can have an international presence with your video content that you could never do before."

For distribution, the logic is the same asset, fractured deliberately. "When you create something on one channel, you want to repackage it for others." A podcast becomes clips, becomes blog posts, becomes LinkedIn posts. He credits the practice with taking his own podcast past 100,000 downloads. RunwayML turns images into video. MidJourney generates imagery — including, when he asks it to "add some melanin in there," something that looks more like him. Canva's data import will merge a spreadsheet of your best-performing quotes and handles into templated graphics at scale. He puts a number on that particular workflow: twenty minutes of work, 30 new followers, 18,000 impressions. And for teams drowning in the whole idea, he points to distribution.ai, which ingests a blog post, podcast or YouTube video and outputs the formats for you.

The objection he anticipates — but is it any good? — he answers with the audience rather than the aesthetics. There are AI-generated YouTube channels with 230,000 subscribers. A history channel he mentions has 27,000.

"People don't care if the content is created with AI. They care whether or not the content is valuable."

Ross Simmonds

One Thing, If You Do Nothing Else

Amid a talk stacked with tools, tactics and screenshots, Simmonds singles out one action and asks the room to do it that night.

He is equally practical about the human side of Reddit access: if a client wants to distribute into a subreddit, his team reaches out to the moderators on their behalf and asks. "Redditors can be nice if you ask nicely."

The Part Where He Predicts Things

Simmonds's forecast is split between the unsettling and the mundane. On the unsettling side: AIs talking to each other; sales development reps sending AI-generated recordings at scale; and — a prediction that lands harder given the synthetic-video demo he just ran — that some of the marketers you're already watching on YouTube are not marketers. "You think it's a marketer, but it's actually just a synthetic person who's pretending that they're real. I think we're going to see a lot of chaos in the industry."

On the mundane side, the future is an errand. You'll tell an AI you want courtside tickets to a Jets game or season tickets to the Eagles, and it will go handle the entire transaction on your behalf. "That's where we're going."

Which is, if you follow the thread, the real argument for everything that came before it. If an agent is going to do the finding, then the question is no longer whether your blog post ranks. It is whether your brand exists — credibly, repeatedly, in the specific places the models actually read.

"While everybody else argues about what is GEO, I want you to start embracing a simple framework that realizes that it's a new dawn. It's a new day. You create valuable content once and you distribute that forever."

Ross Simmonds

He allows himself a shameless plug for his book on distribution, flags it as shameless, and moves on. And then he lands the talk not on a tool or a tactic but on a disposition, which is the only part of the presentation that will still be true after the tool stack turns over.

"We are going through one of the most fundamental changing times in our industry," he says. "If there is one thing that I hope sticks with you, it is this. Have a growth mindset. Keep showing up for events like this. Keep experimenting. Keep tinkering."

Then the closer, which loops all the way back to the opening lie: "And I hope that at least one idea that I shared with you today will stick with you longer than bubblegum."

The applause suggests it might. Whether the budgets move is a different question — the review-site invoices are, after all, still going out — but for twenty-two minutes a room of marketers was handed something better than reassurance. They were handed a receipt.