Design Tokens vs Expression Contracts

Design tokens are named references to design values – color.primary, space.lg, font.heading. They emerged from the need to keep design decisions in sync between design tools and code. When a brand updates its primary color, a token system propagates that change everywhere the token is used. This is genuinely useful. Tokens replaced the nightmare of hardcoded values scattered across a codebase. They introduced a single source of truth for values. They made handoff between design and engineering more tractable.

But a token is a name for a value, not a reason for the value. It says what the color is. It does not say when to reach for it, or whether this particular surface calls for emphasis or restraint, or how this value relates to the surrounding context.

An expression contract is a structured representation of brand intent that an agent can reason about at the moment of design decision. Where a token says what the primary color is, a contract says what the primary color means – the contexts in which it carries weight, the situations in which it should yield, the relationship between it and the brand's voice in this type of content.

A contract is not a constraint list. It is an encoded form of judgment – the kind of judgment a senior designer carries in their head after years of working with a brand, made available to an agent that has no such history.

Expression contracts do not replace tokens. They operate above them. A contract reasons about which tokens to reach for; the tokens deliver the values those tokens hold. The two are complementary, not competing. But only one of them gives an agent something it can actually reason with.

Give an agent a token system and it can produce brand-consistent output in the narrow sense: correct colors, correct type sizes, correct spacing values. Give it nothing beyond that and it will still make the wrong choices – the wrong hierarchy for this content type, the wrong emphasis for this emotional register, the wrong density for this reading context.

The agent was not wrong about the values. It was wrong about the judgment.

This is the gap expression contracts fill. An agent equipped with expression contracts does not just know what the brand looks like. It knows what the brand means to look like in this context – and can produce output that feels considered rather than merely correct.

The token system is the foundation. The contract is what makes the foundation usable for decisions, not just reference. See also: what is LESS Studio, which is where brands author and manage their expression contracts.

Tokens are a dictionary. Expression contracts are the grammar. A dictionary tells you what words exist. Grammar tells you how to use them – which constructions feel natural, which register fits which context, what the rules are and when skilled speakers bend them.

Design tokens have done remarkable work as dictionaries. They gave the industry a shared vocabulary and a reliable reference for values. Expression contracts are the grammar layer that makes that vocabulary productive for agents – not just readable, but usable, not just correct, but purposeful.

The field that needed tokens in the 2010s needs contracts in the 2020s. The problem changed when the composer changed. Tokens were built for engineers. Contracts are built for agents. See also: what AI agents need from design.