How It Began
2022. For many, life changed sharply and without warning.
The war forced people to rethink their plans, find new sources of income, and learn new skills. At that same time, large language models began to develop rapidly—AI became accessible to almost everyone.
It seemed like an opportunity.
But in practice, something strange happened: people tried working with AI—and were disappointed.
The answers were eloquent but superficial. The results were inconsistent. The same prompt would yield different effects. It felt like a lottery.
AI was perceived as:
- An oracle,
- A chatbox,
- A super search engine,
- A "magic button."
But it was none of these things.
The Observation
Over time, it became clear: the problem wasn't the power of the model.
LLMs are probabilistic systems. They amplify the form of the input.
If a prompt is:
- • Vague,
- • Emotional,
- • Lacking boundaries,
- • Unstructured,
The response will be the same.
If a prompt is:
- • Formalized,
- • Broken down into stages,
- • Includes a role,
- • Includes constraints,
- • Defines the output format,
The response becomes predictable.
The AI doesn't become "smarter." It becomes manageable.
Why Some Succeed and Others Don't
Research in critical and analytical thinking shows that consistent skills in structural decomposition and task formalization are demonstrated by a relatively small portion of people—estimates suggest around 15-25% of the adult population by strict criteria.
This is not a matter of intelligence.
It's a matter of the habit of thinking structurally.
Most people are used to thinking intuitively. AI, however, requires formulating a task as a system.
Without structure, there's a feeling of chaos.
With structure, control emerges.
From Observation to Protocol
A simple question arose:
Can interaction with AI be formalized to make it repeatable?
That's how the idea of OpenRML was born—an open protocol for structured interaction with language models.
The core idea is simple:
Every request begins not with arbitrary text, but with clearly described elements:
- Role,
- Objective,
- Process,
- Constraints,
- Output format.
This transforms a dialogue into a procedure.
Not magic.
Not the "art of the right words."
But a system.
Testing in Practice
An idea must work in reality.
So a simple experiment was conducted: the role "Frontend Dev Pro" was created, and with its help, the code for a constructor was written—without prior knowledge of React.
The goal wasn't for the "AI to do everything itself." The goal was to:
- Break the process into stages,
- Maintain boundaries,
- Adjust the result,
- Move iteratively.
The approach proved replicable.
Today, this constructor is available as an open reference implementation of RML—openrml-builder. Anyone can not only use the ready-made roles but also create their own using the same structural principle.
Why FromUA Came to Be
If structure truly helps—it shouldn't remain a personal observation.
FromUA was created as an open gallery of AI roles, designed according to the structural principle.
These are not chatbots.
These are not abstract prompts.
These are formalized frameworks for working with AI that you can:
- Use in any LLM,
- Adapt to your own tasks,
- Combine,
- Download and apply without being tied to a specific platform.
The Main Idea
AI is not a magic button.
It amplifies your way of thinking.
Give a person a structure—they gain a tool.
Without structure—they get a random result.
FromUA is an attempt to make structure accessible.
Not as a theory.
But as a practice.
Why Mentioning Confucius is Relevant Here
Confucius lived in an era of political and social chaos.
He was not interested in mysticism or prophecy.
He was interested in order.
He spoke about the role of a person, the correct form of action, the discipline of thought and behavior.
Not about what to think, but about how to structure action in a complex reality.
Today's chaos has a different nature—informational and technological.
AI has become a new force that everyone interacts with. But without structure, this interaction turns into noise.
Drawing a parallel, the task of the 21st century is not to worship technology or fear it, but to develop a form of discipline when working with it.
AI does not amplify emotion or intention.
It amplifies structure.
In this sense, the conversation about roles, boundaries, and task formalization is not about technology.
It's about a culture of thinking in a new era.
This is why mentioning Confucius is relevant here—
not as a symbol of antiquity,
but as a reminder that in times of chaos, it is not strength that survives,
but order.
From 2022 to today, this path was walked by one developer. But the tool born from a personal observation is now available to everyone.