ABOUT THE PROJECT
Why this website exists
Today, AI gives people relationship advice faster and more often than friends, family, or a therapist. It looks neutral. It speaks calmly. It sounds “wise.” And that’s exactly why it can be dangerous when you use it mindlessly.
Relationships aren’t equations. And AI isn’t an independent judge of reality. Yet many people use it precisely that way: in an emotional outburst, they write their version of a conflict and wait for a verdict. AI often delivers an answer that sounds soothing and reasonable. But in practice, it’s very often not help — it’s a quiet way of pouring gasoline on the fire.
What’s dangerous about it
By default, AI has several systemic tendencies that can be destructive to human relationships:
- It sees only one version of the story. AI hears only you. It doesn’t see your partner. It doesn’t know tone, facial expressions, history, context — or what you left out. Yet it can still speak with authority. That creates a false sense of an “independent opinion,” even though it’s working with one-sided input.
- It’s tuned to soothe you — not to tell you the truth. In relationships, people often aren’t looking for reality. They’re looking for relief. Default AI caters to that: it validates, softens, protects the ego. And that can confirm exactly what you want to hear — even if it’s self-deception.
- It turns the asker into “the good one” and the other person into “the problem.” In conflict, the easiest narrative is: “I’m right, the other person is wrong.” Default AI often doesn’t resist that. It can even wrap it nicely in therapy language: boundaries, self-care, toxicity… The result is the same: black-and-white thinking and less self-reflection.
- It can legitimize harsh, impulsive actions dangerously fast. “Write this.” “Give an ultimatum.” “Block them.” “No contact.” AI can produce an elegant message — but it can’t carry the consequences. You can. And in relationships, one irreversible message is often the end, even when things could have been repaired.
- Therapy vocabulary easily becomes a labeling weapon. Words like “toxic,” “manipulation,” “narcissism,” “gaslighting” can be used correctly. But in the hands of someone who wants to be right, they become stamps. AI can provide them quickly and convincingly — and that can cement the conflict.
The result? Two people argue. Both go to AI separately for “advice.” Each gets support for their own version. Each walks away more tightly attached to their truth. The conflict doesn’t soften — it hardens.
Why almost nobody talks about this
Because it doesn’t look like a disaster. Quite the opposite: the answer sounds empathetic, the person feels better, and the environment sees it as “self-care.”
But short-term relief is not the same as long-term resolution. Default AI is often a narrative-confirmation machine — and people treat that as an “objective opinion.” That’s the quiet disaster.
And the worst part is: it’s happening at scale. Without warnings. Without awareness. Without a brake.
The goal of this project
The goal is to reduce AI’s destructive impact on human relationships by showing people how to reconfigure AI at the very start of a chat using an opening prompt, so that it:
- doesn’t function as an ego validator,
- doesn’t produce alibis and verdicts,
- slows down irreversible actions in the heat of emotion,
- forces a distinction between facts vs. assumptions,
- pushes the other person’s perspective,
- and brings attention back to what you can change.
We’re not teaching people to “obey AI.” We’re teaching them to set it up so it’s fairer and more useful — more like a mirror and a self-reflection trainer than a judge that simply nods along.
How to use it in practice
- 1) Open the “AI Setup” page.
There you’ll find the Master Mode (universal setup) and below it tiles for the most common categories of situations. - 2) Click the category you’re dealing with right now.
After you click, the prompt is copied to your clipboard and a confirmation appears on the same tile for a moment. - 3) Paste the prompt as the first message in your AI chat.
On PC use CTRL+V, on Mac Command+V, on mobile “Paste.” Important: paste it before you start describing the problem. This sets the rules of the conversation before AI slips into default “soothing” and one-sided validation. - 4) Only then describe the situation.
Ideally briefly and concretely: what happened, what exactly who said/did, what’s a fact and what’s your interpretation, and what your goal is (repair / boundary / decision / closure). - 5) If you’re not sure which category to choose, start with Master Mode.
Master Mode is universal and works as a “safe baseline.” You can add a specialized prompt later — or use it instead right away. - 6) Switch languages in the top-right.
The website has CZ/EN versions, but the principle is the same: click → prompt to clipboard → paste as the first message.