When ChatGPT, YouTube and social networks turn customers... into fake experts
There was a time, not so long ago, when visiting an agency was above all a matter of trust. You came for a look. A reading of the field. Expertise built through experience, hindsight, trial and error.
That era is disappearing.
Not because of a technological revolution.
Under the effect of a change in posture.
In just a few months, an entire generation of entrepreneurs has turned to artificial intelligence like handing a microphone to an oracle.
Not to learn.
To validate what they already want to believe.
A well-turned prompt, a voice-over YouTube video, a brightly colored LinkedIn carousel, and here come the answers. Brilliant. Fluid. Ready to use.
📲 "How to earn €10,000 a month with an AI automation agency."
📲 "The perfect sales tunnel in 3 steps thanks to ChatGPT."
📲 "Top 10 prompts for scaling a business without a team."
And when these people come into the agency, it's no longer to ask a question.
It's to test alignment with what we've already told them.
💬 "I warn you, I asked ChatGPT before I came."
And here, the agency knows that the discussion doesn't start from scratch.
It begins with an answer that needs to be deconstructed.
The emergence of the AI customer-expert
At Smart Impact, we support projects in three dimensions that are more intertwined than overlapping: marketing, communication and development. We're a 360 agency, and in our day-to-day work, that means juggling multi-channel strategies, business code, high-value content, and sometimes shaky technical stacks that need to be consolidated without breaking everything.
We're used to vague briefs, shifting objectives, urgent requests and long visions. That's not a problem, it's what we do. What we can see changing, however, is the customer's attitude towards expertise.
Today, we don't ask what we think.
We ask why we don't think like ChatGPT.
Can you feel the nuance?
It's no longer a request. It's a test.
Knowledge without understanding
We're not criticizing the use of artificial intelligence. We use it every day. It's part of our toolbox. We integrate it into our processes, we explore it, we question it. It's not a gadget, it's a lever.
But using AI and blindly following it are not the same thing.
ChatGPT gives structured, clean, reassuring answers. It rephrases with style, aligning concepts brilliantly. It gives the illusion of reasoning, when it's more often than not a linguistic statistic dressed up as the truth.
And therein lies the trap.
It's not the answer that's wrong.
It's the context that's missing.
You ask a question. You get a brilliant sentence.
But we don't know where it comes from, what it's based on, whether it stands up to reality.
We get an opinion.
We think we've got it.
But in reality, we've just read a nice rephrasing of something we haven't mastered.
You can't become a strategist by reading a prompt.
It's not a criticism. It's an honest reminder.
Yes, we can learn. Yes, we can explore. In fact, it's essential.
But no, you can't turn yourself into an expert just because you've got a well-formulated answer.
Between a tip generated in three seconds and a viable, deployable, secure solution that can be integrated into a real environment...
There's a whole world out there.
This is our world. The one we walk through every day.
And that we sometimes have to defend, in the face of an answer that doesn't hold water.
But to the customer, it seemed perfect.
✨ It shines, but it breaks.
And we're here to keep it that way.
Expertise can't be improvised
An agency isn't there to tick the boxes of a plan validated by an AI. It's there to ask the right questions. To confront ideas with reality. To translate an intention into a concrete, sustainable, controlled solution.
And this job can't be summed up in a prompt. It's not just about knowing what to do. It's about understanding why to do it. When to do it. How to do it. With what limits. And for whom.
Expertise is not a compilation of answers.
It's an ability to read between the lines.
To feel when things are going wrong.
To choose when to give up.
We don't call in an agency to validate an idea.
We call on her to make it grow, or to bury it if it doesn't.
And ChatGPT will never tell.
We work with reality
In a project, there's theory, and then there's everything else.
💣 The server that crashes during a campaign
💣 The CRM that changes APIs without warning
💣 The tunnel that stops converting for no reason at all
💣 The budget that needs reframing at the last minute
That's where we work.
Behind the scenes. In the blind spots. In the gray areas.
We know the business.
Not just the agency's.
That of the entrepreneur, the e-tailer, the restaurateur, the consultant, the CEO who is trying to stand on his own two feet by managing a brand, a team, a product and its image.
And no model can simulate that.
🤖 The AI makes shortcuts.
🏗️ The real requires foundations.
You can't turn a donkey into a racehorse
It's a tough formula.
But it is necessary.
The illusion that anyone can do anything with the right tools has become a collective fantasy. As if competence could be born from a well-thought-out copy-paste. As if all you needed was a good tool to become good at something.
But a tool doesn't create intention.
Let alone understanding.
Today, a certain fringe of customers think ChatGPT has made them brilliant.
But repeating an answer is not the same as understanding it.
And knowing a method is not enough to master it.
Being brilliant doesn't mean saying "you have to do funnel automation".
It's knowing when not to do it.
And why.
What's next?
So what do we do with all this?
We could get angry. Send customers back to their prompts. Let ChatGPT drive the projects. But that would be to miss the point of our role. Because our job isn't just to do things for you. It's about explaining, framing and translating complexity into something actionable.
Yes, we can work with AI. Yes, we can integrate automation. Yes, we can support the implementation of an idea conceived upstream.
But on one condition.
👉 Agreeing to put the plan to the test of reality
👉 Not ego-proof
👉 Putting feasibility to the test
Otherwise, it won't be a project.
It's an illusion that costs time, money and credibility.
Because in the end, it's the customer who pays.
And AI won't be there to help manage the breakage.
Finally, AI is not an oracle
ChatGPT is an incredible tool.
But it's not a strategic partner.
He's not an architect. He's not a marketing director.
He's not a developer. It's not a project manager. It's not a human being.
It's a synthesis engine.
Questions for ChatGPT? You're welcome.
But first, ask yourself this:
am I able to evaluate the answer?
Because if you don't, we'll just go ahead with a brilliant sentence...
And no idea where it leads.
So yes, AI can help us go faster.
But if you don't know where to go... you're heading for the wall.
Unfortunately, we're seeing this more and more.
💬 Working with an agency? It means coming up with ideas
🧠 Not with certainty
👣 And above all, leave a little room for experience. She still has a couple of things to pass on
Co-founder of Smart Impact.Passionate about the web from the outset, he launched his first project in 2006: an online music magazine that is still running today. With almost 20 years' experience in SEO, a federal diploma in marketing and a solid geek culture, he and his team transform customers' (sometimes vague) ideas into concrete digital projects.