Skip to content

Introduction

modelARch is a product-modeling and software-generation environment. You describe the product as structured intent, refine its domain and experience, and generate a working implementation from that model.

The central idea is simple: the implementation should remain derivable from the product’s intent instead of becoming a second, increasingly implicit source of truth.

Most generative tools optimize the first delivery. They turn a prompt or specification into code, then hand that code to a team to maintain normally. The first result may be fast, but every later change must recover intent from a growing implementation and preserve decisions scattered across it.

modelARch optimizes the sequence of changes. It keeps a structured, versioned model that captures:

  • the domain and its vocabulary;
  • actors, permissions, and ownership boundaries;
  • actions, events, rules, and exceptional cases;
  • journeys, views, navigation, and interaction decisions;
  • brand and design-system intent;
  • integrations, add-ons, and runtime requirements.

That model is validated and compiled by a versioned generator. A new product decision produces a new model revision and a new release candidate—not an invisible patch to old generated output.

Layer Purpose Examples
Evolutionary Model Author product intent Objects, permissions, rules, views, integrations
Generator Translate supported intent Backend, frontend, tests, contracts, manifests
Runtime Operate live state Data, configuration, infrastructure, telemetry

These layers cooperate without pretending to be the same thing. The model does not contain live database rows. Infrastructure does not decide whether a field was renamed or replaced. Generated code does not become a hidden authoring surface.

modelARch is not:

  • a one-shot prompt-to-code wrapper;
  • a visual editor that merely exports scaffolding;
  • a bidirectional model/code synchronization system;
  • an attempt to place every kind of production state into one file;
  • a claim that generated software can operate without migrations, configuration, infrastructure, or observability.

It is an attempt to make product intent explicit enough that implementation can be regenerated safely throughout the product’s life.

The terms are related but distinct:

  • Regenerative Software is the paradigm and its operating contract.
  • modelARch is an implementation of that paradigm: the authoring environment, model, validation system, generators, preview, and delivery workflow.

This documentation labels normative requirements as such and keeps current product support in the reference sections. That distinction matters: a sound idea should be testable independently of marketing, and a shipped feature should be described precisely rather than implied.