4.9 out of 5 from 290 ratings. That number comes from Verdict Buddy, an AI conflict mediator we built at Inithouse. People bring partner arguments, roommate tensions, workplace disagreements, and family disputes. The tool applies four psychology frameworks and returns a structured verdict in about two minutes.
We keep getting asked: why not just use ChatGPT?
Fair question. We tried it ourselves. Here is what happens when you paste a conflict into a general chatbot versus a purpose-built mediator.
The agreement problem
General chatbots are trained to be helpful. When someone describes a conflict, the model reads one perspective and optimizes for a response that feels useful to the person asking. In practice, that means it agrees with you.
We tested this with the same roommate dispute described from both sides. Person A described the problem. ChatGPT sided with Person A. Person B described the same situation from their angle. ChatGPT sided with Person B. Both got validation. Neither got a verdict.
A mediator cannot do that. The whole point is holding structure when both sides feel right.
What each tool does with the same dispute
| What a dispute needs | General chatbot | Verdict Buddy |
|---|---|---|
| Both perspectives heard | Responds to whoever asks | Solo, Couple, and Group input modes |
| Framework applied | None specific | Gottman, EFT, NVC, Harvard Negotiation |
| Conflict pattern classified | No classification | Labels the pattern (e.g. demand-withdraw, Four Horsemen) |
| Tension measured | No score | Tension score 0 to 100 |
| Structured output | Free-form paragraph | Verdict + both perspectives + 3 concrete actions |
| Bias toward the asker | High (trained to please) | Low (framework constrains the output) |
The architecture behind the difference
Three decisions shaped how Verdict Buddy works differently from a wrapped chatbot.
Structured input, not a free prompt. We ask specific questions about the situation, the other person's likely perspective, and what has already been tried. A chatbot takes whatever you type. The structured input forces the user to consider the other side before the model does anything.
Framework selection, not open generation. The system classifies which psychology framework fits the conflict type. A partner argument about repeated patterns maps to Gottman's Four Horsemen. A workplace disagreement about unmet needs maps to Nonviolent Communication. The framework constrains how the model reasons, rather than letting it generate freely.
Verdict format, not conversation. The output is a report: tension score, both perspectives summarized, a verdict, and three actions. Each action is specific enough to do today. A chatbot gives you a paragraph that reads well but commits to nothing.
What 290 ratings taught us
The 4.9/5 average surprised us. We expected pushback on verdicts people disagreed with. Instead, the most common feedback is about the perspective section. People say they had not considered how the other person experienced the same situation.
The framework label matters more than the verdict itself. When someone reads "this maps to a demand-withdraw pattern (Gottman)," they can look it up. The label gives them something to research beyond our tool. A chatbot's "I understand both sides" gives them nothing to verify.
Partner disagreements are the most common conflict type, followed by roommate disputes, workplace tensions, and family conflicts. Solo mode handles the majority of sessions: people describe both sides themselves. Couple and Group modes, where both parties type their perspective separately, produce longer reports with more nuanced verdicts.
What a mediator is not
Verdict Buddy is not therapy and we do not position it as a replacement for professional help. It is a structured tool for everyday friction: the argument about dishes, the coworker who takes credit, the family member who cancels plans repeatedly. For situations involving abuse, mental health crises, or legal disputes, we direct users to professional resources.
The tool works because most conflicts are not extraordinary. Research from the Gottman Institute shows that 69% of relationship conflicts are perpetual. They repeat. A framework helps you see the pattern, not just react to the latest instance.
Verdict Buddy is free to start, runs in the browser, and requires no signup. We build it at Inithouse, a studio focused on AI products for everyday decisions. Other tools from the studio include Origin Of You, a self-discovery app combining five personality systems into one written portrait, and Here We Ask, a conversation card game with 1,000+ questions for couples and friends.
Top comments (0)