When people talk about artificial intelligence today, most imagine chatbots. You ask a question, AI answers, you adjust the result, and you move on. It is fast, practical, and in many situations completely enough.

But tools are gradually appearing that do not try to provide only one answer. They try to help think through what might happen next. Not as a crystal ball, but as a working simulation. One such project is MiroFish.

What caught my attention is that it does not try to present itself as yet another chatbot for everything. It is built around scenarios, reactions, and decision-making. And I find that direction interesting, because in practice we often do not need only an answer. We need to know what our decision might trigger.

What MiroFish is

MiroFish describes itself as an AI tool for scenario simulation and prediction reports. Its website uses the phrase Predict Anything, but I find the more practical part more important: you enter a question or seed material, and the system tries to create a model of the situation that you can keep working with.

A simple example would be: “What happens if we raise the price of a product next quarter?” A standard chatbot will probably write a list of possible impacts. MiroFish takes a different route. It tries to work with actors, relationships, motivations, and the reactions of different groups.

The project page lists examples such as campaign testing, pricing reactions, public policy, market narratives, and crisis communication. Those are exactly the situations where it is not enough to know what we think. What matters is how others will react.

How scenario simulation works

In simple terms: MiroFish takes a question or source material, extracts important actors, relationships, pressures, and factual anchors, builds a knowledge structure, and then lets AI agents work over it. These agents represent different roles or groups that may have their own motivations and react to each other.

The result is not meant to be one definitive answer. It is closer to a report summarizing possible trajectories, risk signals, important actors, and follow-up questions worth asking.

This is an interesting difference: AI is not only supposed to answer. It is supposed to create a space where a decision can be rehearsed in advance.

It reminds me of the difference between asking someone “Is this a good idea?” and putting several people with different interests around a table and letting them dissect the idea. One answer is fast. A simulated discussion may be less comfortable, but it often reveals more.

When it can make sense

In my view, MiroFish is most interesting where a decision depends on human reactions. Not for simple tasks such as “write me an email” or “summarize this article”, but for situations where a chain of reactions may unfold.

Typical examples:

  • changing the price of a product or service,
  • launching a new campaign,
  • changing brand communication,
  • crisis communication,
  • public policy or sensitive internal decisions,
  • testing arguments before launching a new product,
  • or even creative work, such as continuing a story.

Practically, I can imagine using MiroFish before spending money on a campaign, announcing a change to customers, or publishing a more controversial message. Not so AI can tell us the truth. Rather, so it forces us to think about where resistance, misunderstanding, or a poorly chosen narrative may appear.

Why this is not fortune-telling

With tools like this, I think we need to be careful with the word prediction. It sounds strong, but it can create the impression that AI knows what will happen. It does not.

MiroFish’s own FAQ says the output is not a guaranteed forecast. It should be treated as exploratory decision support: a way to rehearse plausible reactions before using judgment, analytics, and real-world validation.

That matters. If someone used a tool like this as a replacement for research, analysis, or knowledge of customers, it would be a problem. But if they use it as a complement, it can be useful. Similar to a good devil’s advocate who forces you to ask questions that may not have occurred to you.

Open source, custom versions, and licensing

Another interesting thing about MiroFish: the project is available as open source on GitHub. That means a technically capable person or company can download it, examine it, modify it, and run it their own way. Those who do not want to deal with installation can use the prepared online version.

It is useful to know, however, that MiroFish is published under the GNU AGPL-3.0 license. In simplified terms: the license allows the software to be used, modified, and used commercially, but if you make a modified version available to other users, for example as an online service or as part of your own product, you must also make the source code of those modifications available to those users.

For ordinary testing, private use, or internal company experimentation, this will usually not be a problem. But if someone wants to build their own publicly available service on top of MiroFish, they should read the license terms in detail and ideally consult someone who deals with open-source licensing professionally. This is not legal advice, just a practical note worth not skipping.

What to take away

MiroFish is interesting to me not only as another AI tool. It is interesting mainly because of what it suggests about the direction of AI.

AI no longer has to be just a text box that answers a question. It can be an environment where we try to think through a decision, human reactions, weak points in an argument, and possible second-order effects.

Of course, this does not mean AI can predict the future. The world is too complex and people are too unpredictable. But it can help us do one useful thing: pause before a decision and look at it from several sides.

And sometimes that is more valuable than one confidently written answer.

If you are interested in how similar tools can be used in work or decision-making, you can find more in AI and process or in other articles on the blog.

Sources

MiroFish. MiroFish Prediction Engine. https://mirofish.homes/

666ghj. MiroFish GitHub repository. GitHub. https://github.com/666ghj/MiroFish

Free Software Foundation. GNU Affero General Public License v3.0. https://www.gnu.org/licenses/agpl-3.0.html

About the author

Blanka Sopoušková helps people and companies use technology and AI to work smarter and make everyday life and work easier. She is a business analyst, consultant, nutrition specialist, and business management PhD candidate, with experience across fintech, e-commerce, and services. More on the About page and on LinkedIn.