This paper presents a platform architecture for PBM (Play by Mail / Play by Message) and online TRPGs. It integrates the player layer, game master layer, Shared Fate Module, AI support layer, operations management layer, multilingual operation layer, and publisher / rights management layer into a unified design. The central idea of this model is to place AI not as an adjudicator, but as a support layer for interpretation, translation, narrative generation, and operational assistance. Player action declarations, judgment contexts defined by a game master, outcome confirmation through dice, cards, official apps, and other external resolvers, and narrative conversion by a Submaster or AI are separated into distinct roles. In particular, this paper defines the Shared Fate Module as a table-neutral resolution layer placed between players and the GM. This module locks judgment conditions, processes dice and cards, receives external resolver results, confirms outcome bands, and commits those records to the audit log. This structure prevents AI, GMs, and operators from arbitrarily modifying confirmed results. The paper also argues that an AI-assisted PBM/TRPG platform may mitigate GM shortages, enable multilingual participation, reactivate older or minor systems, create authorized distribution pathways, reduce incentives for unauthorized data conversion, and provide value as an AI learning environment. The proposed model does not replace PBM/TRPG with AI. Rather, it reconnects player agency, chance, auditability, rights management, and narrative generation. In preparing this paper, interactive AI systems were used as auxiliary tools for organizingstructure, adjusting wording, supporting figure preparation, organizing translation policy,and supporting TeX/PDF typesetting.AI was used to support document structure, expression adjustment, figure instructions,terminology consistency, and technical formatting. However, the design philosophy, con‑ceptual structure, role separation, terminology selection, and final content decisions in thispaper are those of the author.
Paper
The full text of this publication is not hosted on 44B due to licensing.
Read it at OpenAlex