Multi-Persona Game Design Reviewer


Generative Arts and Design Lab

An agent-based AI tool that helps game designers evaluate new game ideas and define their target audience more precisely.

The Multi-Persona Game Design Reviewer is an agent-based AI tool that helps game designers evaluate new game ideas and define their target audience more precisely. Instead of receiving just a single opinion on a game concept, the system simulates different types of players and their perspectives on the design presented.

A developer can, for example, upload a game design document or a description of their game or individual game mechanics. Any number of AI agents, either self-created or automatically generated characters, then analyze the concept from the perspective of different target groups. These include, for example, casual players, hardcore gamers, story-oriented players, competitive players, as well as younger and older players. Each agent evaluates the concept according to its own interests, expectations, and preferences, and provides corresponding feedback.

The particular added value of the system lies in the fact that it considers not only different target groups, but also allows for differing opinions within the same target group. For instance, two casual players might have different priorities: while one prefers simple and relaxed game mechanics, the other places greater value on social interaction or short play sessions. This gives the developer a more realistic and nuanced picture of possible player reactions, allowing them to define their target audience more precisely, even before a large human test group exists.

The system thus functions as a virtual focus group, helping to identify strengths, weaknesses, and potential trade-offs already at an early concept stage. Developers can see which features resonate particularly well with certain player groups and which aspects might need to be adjusted.

In addition, the tool generates a written and graphical summary of all reviews and assesses which target group and user age currently fits the game concept best. It also offers recommendations on whether the intended target audience should be reconsidered, or whether it would make sense to tailor the design more strongly to a specific target group.

Realized by student Raphael Schilke as part of the seminar "Agenten für Alle(s)" ("Agents for Everything") (SS 2026, Generative Arts & Design Lab, HBKsaar), which contributes to the research of the EU Horizon Europe project HAMLET (GA 101178362).

Abbildungen


Zurück