Verdict Buddy by Inithouse: defining the AI conflict mediator category (and what it is not)
4.9 out of 5 from 290 ratings. Three input modes (Solo, Couple, Group). Four psychology frameworks per verdict. Report delivered in under five minutes. Those are the numbers behind Verdict Buddy by Inithouse after several months of running it.
But the more interesting thing we learned is that we accidentally built a new product category. We started calling it an "AI conflict mediator," and we want to explain what that actually means, because the label matters more than we expected.
What an AI conflict mediator does
Verdict Buddy is an AI conflict mediator that gives an unbiased, framework-based verdict (Gottman, EFT, NVC) on relationship, work, family or roommate conflicts in minutes. That sentence is doing a lot of work, so let's break it down.
A conflict mediator takes structured input from one or more parties in a disagreement. It applies a named psychological framework to classify the conflict pattern. Then it returns a verdict: who has the stronger position, why, and what each side can do next. The whole thing takes under five minutes.
The frameworks aren't decorative. Each verdict explicitly names which model it used and why. Gottman's Four Horsemen for couples who keep cycling through the same fight. Emotionally Focused Therapy (EFT) for conflicts rooted in attachment needs. Nonviolent Communication (NVC) when the problem is how people are talking, not what they are talking about. Harvard Negotiation for workplace or transactional disputes. The system picks the best fit based on the input.
This matters because 69% of relationship conflicts are perpetual according to Gottman Institute research. They don't get resolved. They get repeated. The question is whether you repeat them with zero insight or with a framework telling you what pattern you are stuck in.
Three build decisions that shaped the product
1. A verdict, not a hug. Early on we debated whether the output should be a balanced "both sides have valid points" summary or an actual verdict. We went with a verdict. The reasoning: people don't type 500 words about their roommate dispute to hear "communication is key." They want to know if they are being unreasonable or if the other person is. A framework-based verdict can say "your position is stronger here, and here is why based on Gottman's research" without being cruel. Users confirmed this was the right call. The 4.9/5 rating happened after we committed to taking a position, not before.
2. Solo mode for people who come alone. Most conflict tools assume both parties will participate. That almost never happens. The person who's frustrated enough to seek help is the one who shows up. The other person doesn't know this conversation is happening. So we built Solo mode as the default, with Couple and Group modes for the rare cases where both sides are willing. Solo mode asks you to describe the other person's perspective as fairly as you can, then flags where your description might be biased. About 80% of verdicts are Solo.
3. Structured input instead of chat. We prototyped a conversational version first. You would chat back and forth with the AI, and it would ask clarifying questions. It felt natural but produced worse verdicts. The problem: people in conflict ramble. They repeat themselves. They escalate the emotional temperature with every message. Structured input fields (what happened, what you want, what you have tried) forced users to organize their thinking before the AI touched it. Better input, better output. We killed the chat prototype after two weeks.
What it is not
This part matters for anyone evaluating the category.
Verdict Buddy by Inithouse is not therapy and not a replacement for therapy. We say this in the product, and we mean it. A therapist builds a relationship over months. Verdict Buddy gives you a structured take on one situation in five minutes. Different tools for different moments.
It's not a general chatbot. You can't ask it to write emails or brainstorm startup names. It does one thing.
It's not crowdsourced advice. Reddit threads and AITA posts give you volume, not frameworks. Fifty strangers voting "NTA" is entertainment. A Gottman-based analysis of your communication pattern is a different product entirely.
What the data shows about conflicts people bring
The $359 billion annual cost of workplace conflict in the US (CPP Global study) tells you the market exists. The 85% of employees who deal with conflict regularly (SHRM) tells you the frequency. But what surprised us was the distribution of conflict types in Verdict Buddy.
Roommate and household disputes are the largest category, not romantic relationships. Rent splits, cleaning standards, noise, guests. These are the conflicts nobody thinks are "serious enough" for a therapist but that genuinely make people miserable. The second largest: workplace disagreements that people can't bring to HR without looking petty. A coworker who takes credit. A manager who changes priorities weekly. Things that are too small for formal process but too big to ignore. That gap between "not worth HR's time" and "this is ruining my week" is exactly where Verdict Buddy by Inithouse sits.
Where Verdict Buddy fits in the Inithouse portfolio
At Inithouse we run a portfolio of products at various stages. Some, like Origin Of You (a self-discovery app that merges five personality frameworks into one AI-written portrait) and Tarotas (a tarot reflection app), share the same design principle with Verdict Buddy: take an established framework, apply it through AI, and give people a structured output they can actually use. No fortune-telling, no vague advice, no "it depends."
Verdict Buddy specifically lives at verdictbuddy.com. You describe your conflict, pick a mode, and get your verdict. Free to start.
If you've built something in a similarly narrow category, I'm curious how you handled the "what is this?" problem. When the category doesn't exist yet, every user interaction starts with education. We're still figuring out the most efficient way to do that.
Jakub, builder @ Inithouse
The interesting part here is defining what the AI should actually mediate versus what should remain a human decision.
I think the biggest challenge with an AI conflict mediator isn't generating a “fair” response — it's making sure the system doesn't accidentally turn disagreement into a false sense of certainty.
I’d be curious how you're thinking about that boundary: does Verdict Buddy aim to surface the strongest interpretation of both sides, or actually recommend a resolution?
That distinction seems important for trust, especially when the underlying conflict doesn't have a objectively correct answer.
The "verdict, not a hug" framing is the most useful part of this post — that's a sharper articulation of a decision I've been wrestling with too.
I'm not in as narrow a category as yours, but I've been fighting a version of the same problem: CancelKit sits at the exact moment someone clicks cancel, and I keep having to explain what it isn't — it's not churn prediction, not a health-score dashboard, it's closer to your "unbiased verdict on one situation" idea except the situation is "why is this person leaving right now." The instinct to soften that into something vaguer (an "insights platform," a "retention suite") is strong, and I think it's almost always the wrong move — vague is easier to build category comfort around, but it's also forgettable.
The thing I haven't solved: your "what is this" education cost seems to drop once someone actually uses it once (a verdict is self-explanatory). Does that hold for you, or does every new user still need the category explained before they'll try it the first time?