I bet on Flash. Then on apps. Which is why today I bet on the niche

Before the internet arrived commercially in Brazil, I ran a BBS.

For anyone who missed that era: it was a computer plugged into a phone line, waiting for another computer to call it back. One at a time. The person dialled, the modem screeched, and you had a network. That was what existed.

From there I programmed in DOS, wrote in Lotus, watched HTML appear, watched Flash be born and die, watched apps become a national obsession and then become occupied space on a phone. And today I am here discussing how to get a piece of text cited by an artificial intelligence.

Twenty-six years doing this give you something you cannot buy: a ruler. You start sensing that something has shifted before you can prove it.

And they give you one specific scar, which is what I want to talk about here: the difference between the hyped and the niche. Because I bet on both. And what I learned is not what people usually write.

The hype is almost never a lie

That is the first thing almost nobody says.

When we discovered Flash, we thought the world was going to change. And it did. Flash delivered motion, interaction and capability that simply did not exist in HTML at the time. It was not a salesman's story: it was a new, real capability in the hands of anyone willing to learn.

Flash did not die for being bad. It died because the environment changed around it. Browsers adapted one way, it went another. Some accepted it, some did not. Mobile arrived, open standards arrived, the security bill arrived. Adobe announced the end in 2017 and closed it for good on 31 December 2020 — Apple, Google, Microsoft and Mozilla switched it off together, each in its own browser.

Hold on to this sentence, because it applies to everything that came after:

Technology rarely dies for being bad. It dies when the world changes around it.

And then comes the uncomfortable part: you have to bet anyway. Whoever did not look at Flash in 2001 lost years of advantage. You will not always get it right — but betting is how you make a difference. Standing still waiting for the dust to settle is also a decision, and usually the worst one.

The mistake is never betting. It is betting everything.

The bet that cost me most did not cost me money

This is the story I do not usually tell.

There was a time when everybody wanted an app. Shops wanted one, clinics wanted one, service providers wanted one. And I believed it. I invested heavily: knowledge, equipment, team hours, to develop for iOS and Android.

Anyone who never developed has no idea what that was like from the inside. It was not enough to make the app work. It had to be certified, it had to hold up on the iPhone 10, the 11, the 12, the 13, the 15, and on and on. Every new version of the operating system meant rework. The cost of keeping that standing was abusive.

And most of the ideas should not have existed. People showed up wanting an app to track hours — and the phone already has a native clock. When I said it was not smart to build that, the client took offence. They left here annoyed and went to consult an agency, which said "great idea, wonderful" and charged for it.

Meanwhile, the ground was moving. Space on the phone ran out. People started deleting apps instead of downloading them. Concern arrived — about personal data, about photos, about accounts, about what that software was doing in there. Nobody wanted one more icon on the screen.

That was not just my perception. As early as 2014, a comScore measurement in the United States showed that the majority of smartphone owners downloaded exactly zero apps in an average month — around two thirds — and that half of all downloads came from just 7% of users. In other words: at the peak of the fever, the real market for "downloading a new app" was already a small club.

And here is the point that matters. I did not lose money on that one. What came in went back out. I broke even. I learned a lot, and everything you learn is useful.

Even so, if I could go back, I would take part of my focus out of there.

Because the cost of hype almost never shows up on the bank statement. It shows up on the calendar. Two years of attention, of team, of learning and of investment that went into something that evaporated — and did not go into something else that would have lasted. Nobody goes bankrupt because of hype. People break even, feel up to date, and only much later notice what they did not build in the meantime.

Who makes money selling hype

Here I need to be direct, even knowing it is the unpleasant part of the text.

Hype is not sold by the market. It is sold by whoever profits from getting a project approved.

I am not an agency. I never was. I am a creation and development workshop — I work with what I like, and that is why I am still here after all this time. I want to have the solution and to deliver the solution. That changes everything, including the conversation I have with a client.

The most common model out there is another one: rent a floor in a nice building, hire a junior team, set up a structure of account contacts, service leads and assistants, and the client pays several times the value of the work — happily, because the setting justified the price. And most importantly: in there, people almost always say what the client wants to hear. Because disagreeing costs you the contract.

I am not generalising; I know very serious people working that way. But the mechanism exists, and it is the engine of every hype: whoever is paid to approve projects is never going to tell you that project should not exist.

It is the same character who today sells certainty about artificial intelligence to people who have no way to check. Whoever sells a product fires that guy in thirty days, spreadsheet in hand. The dentist keeps paying for two years.

So what is hyped and what is niche, exactly

Forget marketing for a minute and think of yourself as a customer.

You work in one field. But you buy in a completely different one. You play golf, you play tennis, you follow football. Notice what happens when you go looking for information about it.

The generic gives you the rule, the regulation, the fixture list. It is correct, it is useful, and you find exactly the same thing in fifteen places. It has no signature. There is nobody in there.

Now imagine you are choosing a racquet and a ball, and a tennis player shows up to talk to you. Not advertising — talking. Telling you what that racquet is good at, what it is bad at, what changes in your spin, what you will feel in the first week and what only appears in the third month.

It is another universe. And the difference is not information quality: it is signature. There is somebody in there, with a name, with an opinion, with something to lose if they are wrong.

That distinction always existed. What changed was its economics.

Generic content became a commodity at the exact moment the machine started producing it for free, in any quantity, at any hour. Fixture lists, rules, basic explanations: all of that comes out in three seconds today and costs nothing.

What is left is what has an owner. Not because the machine is bad — because the machine does not play tennis.

Hype is a moment. Niche is consistency

There is another difference, and this one shows up in the cash flow.

Hype is a spike of attention. You may well close a sale there, and sometimes you do. But it is borrowed attention: the subject ends, the audience ends.

The niche is people who went looking for that. That person will consume consistently, and you become their supplier. They know where to look. They come back. Because they realised you understand the bitter taste of that thing and know which sweet they want.

It is the whole relationship inverted: hype gives you a customer, niche gives you recurrence. And recurrence is the only thing that accumulates.

I have the proof of that in my own house, and it is slightly embarrassing. On broad searches, the kind for "video production company", Animaker sits on page four in Portuguese — lost in the crowd. On "animation studio", which is more specific and has fewer people searching, we show up near the end of page one. The broad term has far more searches and brings me nobody. The specific one has fewer searches and brings the right people.

The generic competes with everyone. The specific competes with almost no one.

It is the same story as SEO becoming GEO

And this is where twenty-six years are good for something, because you can see the pattern repeating.

Classic SEO was a contest for position in a list. Ten places on page one, and you fought for one. In that game, being broad helped: you appeared in a lot of generic searches, you picked up volume, and volume turned into clicks.

GEO — optimising to appear in artificial intelligence answers — is a contest to be cited in a single answer. There are no ten places. There is one answer, with two or three sources.

That changes the entire premise. In a list, being average leaves you visible. In a single answer, being average is the same as not existing, because the machine needs a reason to choose you over the average — and the only possible reason is that you said something the average did not.

In other words: the niche stopped being a clever strategic choice and became the door. There is still room for the generic, except now it is occupied for free by a machine that never sleeps.

Where I am betting now

I said the mistake is not betting, it is betting everything. So it would be cowardly to end this text without saying where I am putting chips today.

Three things.

A place of my own. A site that is mine, content that is mine, an address that does not change because a platform changed its rules. It is the only asset that accumulates. I have written about that in detail in another piece, and by this point you already understand why.

Content with a signature. It is not about publishing more. It is about publishing what only I can write: the process, the real number, the real timeline, the opinion with an owner, including the part that does not flatter me — like the app story up above.

Artificial intelligence declared, not hidden. We use AI in the operation, and we say where we use it, how we use it and what does not go into it. There is a whole page on the site about that. A handful of production companies do this today; three years from now the ones that do not will have to explain why.

I may be wrong about any of the three. I have been before, and I wrote about it here in the middle. But that is the nature of the game: you bet, you learn, you adjust — and you keep building on what is left.

What I am certain of is the opposite: whoever bets on nothing will arrive in 2030 with exactly what they have today. Which is the only guaranteed way to lose.