After prompting since the early days of AI apps...

Is Today’s AI Really That Far Ahead?

Of course, today’s AI is light years ahead of the early days.

And because progress has been so crazy fast, I am not so sure it is not like something I once learned in computer science about CPU production:

In short, even a tiny distance between the CPU core and its inner supporting memory can make the CPU perform better (way faster) than originally designed. Some of those are set aside for the “next generation,” even though they are already in production, while the lesser models are sold.

Same with AI !!! (in my humble opinion)

It sometimes feels like they already have a lot figured out, but they release it "commercially" when and how they want...

Why Does the Warning Still Sound the Same?

One thing has not changed from the early release until today.

That message.

If I am correct, not even one word changed:

“AI can make mistakes. Check important info.”

By now, it should at least say:

“AI can sometimes make mistakes. Check important info.”

Is It Really the Core AI, or Is It the App Logic?

Well, I am certain that It is not only because of AI technology development.

It is also because the UI of AI apps has forced-logic running in the background, It feels forced, like it’s being shoved down the user’s throat.

Naturally, that logic could be designed per user or user profile. But in reality, it has to be one logic fits all.

And by “logic,” I mean this:

When you search, it is not just the web search tool, the famous citations, the sources, or even quick scraping of source code to verify something.

It is the “mind” of the system deciding what the user’s intent is.

Then it delivers results based on that interpreted intent, together with a pile of fixed factors each AI platform has adopted in a sincere attempt to provide what the user searched for.

This is insane. I doubt they can even count how many angry responses they get from users who ask for A and get B instead.

What About Generative AI?

This logic also applies to generative AI.

And again, it has improved tremendously.

Image AI today is not even comparable to the old days.

But how many times do you still get a crack in the result?

Or, more likely, how many times do you just get pissed off?

So What Happened With SEO, AEO, and AI Visibility?

Over a year ago, at the office we found out that old-school SEO tools like were only patching for AI visibility.

They kept pushing the same keyword and backlink agenda, as if SEO did not take a hit.

I told my colleagues in the office that I 💪😎 was going to do it all with AI.

At that time, I considered myself an AI prompt champion.

And honestly, I was, and I still am 😊 pretty good at it.

I was helping many people get more or less what they were trying to achieve.

But for SEO, AEO, and getting sites more visible on AI and LLM search, nothing worked like it used to.

Not like in old days SEO when you followed something like the Semrush process to the tee.

Did Prompting Fix It? Nope.

After investigating and testing, and I know it sounds like a quick thing, trust me, the nights 🌒 I spent on it were countless.

I could not get any of the top AI UIs to bend to my needs.

They insisted on sticking to the one-logic-fits-all plan.

And since I have worked with all of them from their birth, I know now that this part will never really change.

Actually, the more commercial they get, and very soon once ads come in, it will probably get even worse....

So What Is the Real SEO Problem Now?

Like many colleagues in the AI visibility optimization industry, I started looking for a solution.

My scientific conclusion was this:

SEO remains important. BUT !! people ranking at the top of SERP still get little or no traffic, or their traffic is exponentially reduced. LLMs use Google and Bing SERP to support web search. BUT !! they do not fully trust it. So they scrape website data to verify what is on top of SERP or GBP.

And remember this, because this is the key:

The content has to fit the user logic. Yes, guys, the interpreted user-intent one. But let’s call it the user search intent.

How Much Does Content Matter?

After deep analysis and more tests (🌒🌒🌒) than I can even count, I told my manager my conclusion:

🚨Content, even raw content on a web page, counts for at least 60% to 70% of the weight in whether a site shows up in LLM search results.

The rest is what we already know: Citations, sources, reputation, footprint, and so on.

Can Prompts Handle Content Gaps?

That question led me back into an insane adventure.

I tried to produce prompts to handle the content gap process. But AI pushes its own agenda. It distorts almost everything you try to achieve by prompt.

That applied logic just does not go away.

What Happens When You Finally Get the Report?

👌 Let’s say I finally get the reports I want...

🚧 Then I am stuck with long chats that barely move anymore. Sometimes they stop ✋🏻completely. And I cannot continue the thread.

📚 Or I end up with tons of files in folders on our network.

😵‍💫 And very quickly, we lose track of everything.

So What Is the Better Way?

Then I realized something.

And this is my take:

Prompting is fun. Challenging AI is fun. BUT for this purpose, it is doomed to fail. Not because prompting is useless. It fails because of all the reasons I mentioned, and some more I did not even get into.

To get away from that annoying magnetic AI logic, to actually achieve results, and to avoid getting lost in files, dead-end chats, and stuck threads, it is better to pay for and use a platform !!!

A platform that uses core AI, but applies the content gap analysis logic properly.

Did Old-School SEO Tools Have the Answer?

Not even close, when I looked it up, Semrush, Ahrefs, and others indeed came up with “content gap” searches.

But mostly, it was just an informational 📖 page.

Then they quickly went back to 👵🏻 old-school SEO.

I was looking for next-generation content gap analysis.

A platform focused only on that.

I found one.

And now I am twice as happy.

No more tedious prompting like crazy.

Even better, with a few clicks, I get the results.

Straight to our content writers.