Quoth is built through directed AI development — a disciplined practice in which the product owner drives every decision, and AI serves as the implementation engine. I am not a trained software engineer. I do not write every line of code. But I conceive the structure, define the data model, establish the design language, and make every meaningful product decision before a single line of code is generated.
The AI handles syntax; I handle intent.
The term "vibe coding" was coined in 2025 to describe a way of working in which users generate applications by prompting AI assistants in natural language — rapid, high-level, "vibes"-driven. It is useful for prototypes, small internal tools, and weekend experiments, but it carries a reputation for a reason: code produced that way is often built back-to-front, with structure arriving after the fact, if at all. Bugs hide in places no one has mapped. Foundations are improvised. The app works until it doesn't. That is not how Quoth is being built.
Vibe coding means accepting whatever the AI produces without scrutiny. That is not this.
Every feature in Quoth is preceded by schema analysis, structural review, and a deliberate decision about what should exist, what should not, and why. Planning and design happen in a dedicated environment, separate from code execution. Features are analyzed, scoped, and documented before implementation begins. The sequence is always the same: audit before brief, brief before code, verify before merge. That discipline — separating thinking from building — is intentional, and it matters.
This discipline is written down. The Quoth project standards are a living document of thirteen rules and two process principles that govern every change to the codebase. Files are capped at five hundred lines. Every piece of code must have a declared home. Every data type has exactly one source of truth. Components render, functions think. Names follow conventions that are non-negotiable. TypeScript errors are design feedback, not obstacles. Shared UI lives in shared constants. Stability comes before features. The full list is longer — rules for data modeling, version control, definition of done, and more — but the shape is consistent: decisions are made in advance, written down, and applied to every change.
Every feature decision starts with a simple question: does this serve the person at the table? Quoth is built for GMs in the moment — not for technical showcase, not for engineering elegance as an end in itself.
Does this serve the person at the table?
The images that bring our worlds to life deserve to be treated with care. Tabletop roleplaying games exist, in no small part, because artists made them real — the dragon on the cover, the ranger in the margin, the sprawling city map that made a player lean forward and say, that is where we are. For generations, fantasy artists have done the quiet, essential work of giving form to imagination. In doing so, they did not merely illustrate a hobby. They defined a culture.
Quoth is built in respect for that legacy. When a GM uploads an image, two optional fields are offered: an attribution line and an AI Generated flag. Neither is required. Both are permanent. Credit stays visible with the image for as long as it lives in Quoth. The AI Generated flag is a good-faith tool — Quoth does not prohibit AI art, but it asks GMs to disclose it and displays that disclosure plainly. Voluntary attribution is part of Quoth's approach to responsible image handling and DMCA compliance: a record of stated origin, preserved in context. The conversation about AI-generated imagery in the TTRPG community is real, ongoing, and not simple. Quoth does not take a side — but it does ask for transparency.
The result is an app that was assisted by AI in its construction, but not surrendered to it. Quoth is an independent project, a work in progress, built with care and held to a standard.
— Eric C. Mason, Ph.D.