Aren't we underestimating AI just a little? My personal vision on what's coming

The gap between people who use AI and people who actually build with it is growing every day.

I see it in almost every conversation I have. With entrepreneurs, with managers, with people who call themselves AI enthusiasts. A divide is forming between people who use AI mainly as a search engine or to fix spelling mistakes and people who use it to automate, to build and to pioneer.

I'm in that second group. And what I see makes me both excited and uneasy.

This is my personal vision of where things are headed.

AI is becoming more accessible.

For a long time, the barrier to using AI well was knowledge. You had to know how to write a good prompt. You had to understand what the system could and couldn't do and especially how to get quality output from it.

I see that changing fast.

With the integration of voice, our input is getting easier. You just talk and with speech you naturally give more context than you would by typing. AI understands you, executes, responds. The typing interface is becoming increasingly invisible.

And the models themselves are becoming more accessible too. I saw it clearly with the launch of Claude Fable 5. Earlier models needed explicit instructions. You almost had to force the system in the right direction. Fable 5 reads intent. You give it the destination and it figures out the route. You need to know less and less to get more and more done.

Voice on the outside. Intelligence on the inside. Together they make the barrier to getting quality output almost disappear.

And once that becomes the standard, AI is no longer just for tech people or early adopters.

My mother. Your father. The owner of the bakery around the corner. They will all start generating increasingly powerful output.

The role of people.

The first wave was about individual tasks. Writing a text. Answering a question. You give the instruction, AI executes.

What we are working with now is already different.

At Twenty Twenty Solutions we build a lot of agentic AI. Agents that work independently, not because you instruct them every time, but because you told them once what you want. Customer service. Invoicing. Follow-up. The entire process, without human intervention.

The execution layer is moving to a system. The admin employee processing invoices. The customer service rep answering the same ten questions. These are not distant future scenarios. This is what is already happening.

What remains for people is the controlling role. Monitoring. Adjusting. Deciding what the system cannot yet decide. That will become the most valuable position, but only for people who understand how these systems work.

The shift is honestly harder than most people realise. Not everyone will move with it. The ones who do are the people who are already experimenting and trying to understand what is happening under the hood.

A new target audience, from human to AI.

Since the beginning of the economy we have focused on one thing: the human being.

Every market, every product, every business model has been built around human needs and human behaviour. The merchant in 1400. The advertiser in 1980. The UX designer in 2010. All the same principle: understand the human, reach the human, convince the human.

That is going to change.

We are building software for AI. An API that enables an agent to work faster, more efficiently or more cheaply. No buttons for a user. No screens to navigate through. No onboarding for a human who needs to learn how to use it.

A concrete example: imagine you build a tool that helps companies schedule job interviews. Today you build that with a beautiful interface so a recruiter can navigate through it easily. Soon you will build an API so an AI agent handles that scheduling autonomously, without a human ever looking at it.

Same problem. Completely different target audience.

With a background in marketing and advertising, I think this might be the most fascinating and at the same time most underestimated development right now. Your target audience is shifting from human to system and that requires a completely different way of thinking about building, selling and communicating.

Economy for AI.

When AI agents operate autonomously, retrieving data, purchasing services, hiring other agents, they need a way to pay. Without a passport. Without a bank account. Without a human approving a transaction.

In the short term they will probably choose the most efficient route that already exists. Decentralised, borderless, without intermediaries. Crypto makes sense, not because AI has a preference for it, but simply because it works better than fiat for automated systems.

But the question that really occupies me goes further than that.

What if AI eventually starts determining what value actually is?

We humans invented currency because we needed trust between people. Gold had value because we collectively decided it did. The dollar has value because a government guarantees it. But all those constructions are built on human agreements, human trust and human institutions.

An AI system has no use for any of that.

If AI can assign value itself, to data, to computing power, to information, to access, why would it choose a system we designed for our own needs? Maybe an entirely new form of value exchange will emerge that we cannot yet comprehend.

That is not something that happens tomorrow. But the direction is inevitable. AI provides the intelligence. Blockchain provides the financial layer for now. But the final destination? The system will determine that itself.

The development does not wait.

I don't know how quickly these things will actually happen.

What I do know is what I see every day. The entrepreneur who was sceptical about using AI in their organisation last month is now using it every day. The team that was manually running processes six months ago has now fully automated those processes. The tools we build today already feel outdated the next day, that is how fast it goes.

Everything we build today is already a step behind tomorrow. That is not a problem, that is the reality we operate in. And the people who accept that and keep moving will build an advantage that will soon be almost impossible to close.