Glossary

What Is an Architecture Design Document (ADD)?

TL;DR: An Architecture Design Document (ADD) is a technical blueprint describing the overall structure of a software system. It is written before development starts and provides everything a software developer or an AI coding agent needs to make consistent decisions across the entire build.

What it contains

A complete ADD establishes the project's foundation. It typically covers:

  • System overview: What the app does, who it is for, and what it explicitly will not do.
  • Data schema: Tables, models, relationships, and how data flows between them.
  • Tech stack: The specific frameworks, libraries, and third-party services in use.
  • Routing and component structure: How the app is organized and how the pieces connect.
  • Boundaries: What the developer (or agent) is absolutely not allowed to change autonomously.
  • Phase sequencing: The exact order in which features should be built, organized by dependency.

How it differs from a PRD

A PRD describes what the user experiences. An ADD describes how the system works internally. The PRD is written for the product and design team, while the ADD is written for an engineering team, whether that includes human developers or AI agents doing the building.

Why it matters for AI coding agents

When you prompt an AI coding agent without an ADD, the agent infers your architecture from context: the files it can see, the errors it encounters, and generic patterns from its training data. This works for small tasks, but as complexity grows, the agent begins making contradictory decisions and loses track of your earlier constraints.

An ADD gives the agent a stable source of truth to anchor against during every session. It is the difference between an agent that executes your architecture and one that invents its own.

Zalcro generates an ADD as the first output of every planning session, paired directly with a system diagram and a phased implementation plan.

Related terms

Used in