A component intent library for helping humans and AI tools understand reusable patterns.
- Year
- 2025
- Focus
- Component intent for humans and AI
- Role
- Design systems, documentation design, component strategy, AI workflow exploration
- Themes
- Design systems, Documentation, AI workflows, Component strategy
How can a design system explain why a pattern exists, not only how it looks?
Case specimen
Component intent for humans and AI
- Messy inputs
- Component variants · Usage rules · Anti-patterns · Design intent
- Working frame
- Document components as reusable decisions with purpose, constraints, examples, and boundaries.
- Decision enabled
- Help designers, engineers, and AI tools choose the right pattern and avoid plausible but incorrect substitutions.
- What changed
- The lab reframed design system documentation as shared product reasoning rather than a static component catalogue.
The situation
Context
Design systems are increasingly used across tools, teams, and automated workflows. Documentation has to support people and machine-assisted work without losing design judgement.
Problem
Component libraries can become visually complete but conceptually thin. Without intent, teams reuse patterns inconsistently and AI tools generate convincing but inappropriate interfaces.
The work
My role
I explored component documentation structures, wrote intent patterns, and prototyped how usage guidance could support both human readers and AI-assisted generation.
What made it hard
The challenge was adding meaning without creating documentation that felt too heavy to maintain.
Process
I audited common component docs, identified missing judgement layers, and shaped a repeatable structure for purpose, anatomy, behaviour, constraints, and examples.
Key design decisions
The model emphasised plain-language intent, misuse cases, decision prompts, and constraints that could travel across design files, docs, and implementation prompts.
What it changed
Outcome
The concept gave the design system a clearer role as a source of product reasoning, not only a source of reusable UI.
What I learned
A component is not just a shape. It is a decision with context, tradeoffs, and boundaries.
What I would do differently
I would connect the intent model to real code examples and evaluate whether it improves generated interface quality.