Yuji Horii has spent decades making games people remember. His latest experiment is considerably smaller: a free browser mystery that he says took about 30 minutes to put together with AI. The Dragon Quest creator shared it on October 11, and today's English-language coverage has already turned a modest personal project into another argument about what counts as making a game.

The timer makes an easy headline. Horii's explanation is more revealing. He compared the experience with making The Portopia Serial Murder Case alone, when he had to handle the story, pictures and programming himself. What excited him this time was being able to turn instructions into something playable without personally doing every job along the way.

What He Actually Released

The game is called Interrogation Room Komiya, or 取調室の小宮. Famitsu's report describes a compact interrogation adventure: you're a police officer questioning a 72-year-old former locksmith suspected of a pawnshop safe robbery. His statements don't always match the evidence, and you have to use those contradictions to break down his account.

There are two emotional parameters to manage, agitation and resistance. Push resistance to 100 and he stops talking, so uncovering a lie isn't simply a matter of hammering the most aggressive question available. The reported setup gives the player something to weigh: keep pressing for answers, but avoid losing the conversation altogether. Whether that holds up in play is a review question. The premise itself is at least specific enough to discuss beyond the tool used to build it.

The Portopia Comparison Is Doing Some Work Here

Horii wasn't comparing a half-hour experiment with the production of a modern Dragon Quest. He was remembering the practical burden of working alone. The distinction matters, particularly for anyone who has a game idea but can't comfortably handle the art, programming and writing required to realise it.

There is also an older piece of context worth bringing back. In Famitsu's December 2025 account of his G-STAR appearance, Horii described enjoying the process of thinking up games while finding the work of implementing them difficult, including building and processing large amounts of data. That doesn't tell us exactly how he directed this new project. It does help explain why reducing the distance between an idea and a playable version might appeal to him.

Reading his latest post as a claim that game design has become effortless misses that context. The attraction he's describing is being able to attempt something himself, quickly. Plenty of small developers will recognise the appeal even if they have no interest in letting a model anywhere near their final artwork or dialogue.

The Verdict Started Arriving Before Much Discussion of the Game

GamesRadar+'s headline frames the release as Horii continuing his “love affair with AI slop”. The article puts quotation marks around game and moves quickly towards concern about whether his enthusiasm could carry into Dragon Quest XII. That's a fairly loaded way to introduce a small browser experiment.

Disliking generative AI is a reasonable position. So is asking how material was generated, what rights sit behind it and whether the result is worth playing. But labelling a project slop doesn't answer those questions. If the interrogation is shallow, the writing incoherent or the choices meaningless, describe that. Those would be useful criticisms of this particular game, rather than a verdict borrowed from the method of production.

We saw a related divide in Mermade's discussion around AI-assisted coding, where the label itself became a reason some people would refuse to play. Horii's experiment puts a familiar developer's name into a conversation smaller creators have already been having.

Thirty Minutes Still Needs to Be Kept in Perspective

The build time is Horii's own estimate. We haven't independently verified it, and a small interrogation game doesn't establish how long a larger project would take, how much revision it would need or what it would cost to support. The reports also don't settle how much of the scenario and wording came from Horii versus the model. Those details matter when making claims about authorship, not just speed.

His reputation doesn't automatically make this experiment good. Equally, using AI doesn't tell us that his judgement stopped contributing. The useful questions are what he asked the tool to produce, what he chose to keep and whether the resulting decisions make an interesting game. A rapid build is a reason to look more closely, rather than skip that assessment.

For solo developers, that is the part worth following. A smaller practical barrier could let someone test an idea they would otherwise leave in a notebook. Getting that first version working is valuable. Finding out whether anybody wants to keep playing it remains the next job.

Sourcing note: this article draws on Horii's October 11 statement as reproduced by VGC and Famitsu, Famitsu's descriptions of the game and his earlier G-STAR discussion, and GamesRadar+'s October 12 coverage. This is news analysis, not a hands-on review; the production time remains Horii's own claim.

Would you judge Horii's experiment differently knowing how it was made, or would the interrogation itself decide whether you kept playing? Let us know your thoughts in the comments. And if you've got a game you'd like us to review, send it over to editor@gamevibe.io.