Intent Is the Interface

Tagline: "The screen was a constraint we mistook for the product."

Summary

Ramon Marc argues that designers have been mistakenly focusing on screens (interfaces) rather than the underlying user capabilities and intents those screens represent. With the rise of AI, voice, wearables, and agent-driven interactions, this screen-first approach is breaking down. He proposes a paradigm shift: design around intents and capabilities, letting interfaces be derived dynamically from context.

Key Points

We're designing the wrong unit.

An interface is merely an "appearance" of something deeper: a capability — "the thing users can actually do." "Request a ride" is the capability; the app screen, voice command, and watch tap are each just different appearances of that same capability.

"Decades refining shadows. The object casting them got neglected."

Capabilities are likened to cells in a petri dish: "They exist before you observe them."

Capabilities decompose into intents.

Auditing any interface reveals actions map to a small set of patterns — the author proposes eight intents that cover most of what we build. Uber's "Request Ride" is broken into three intents, shown as three projections (phone, watch, voice).

Context is introduced: send during calm vs. send during crisis produces different UI expressions — "Same intent. Different situation. Different expression."

Context has at least six dimensions, of which screen width is just one.

Core claim: "The interface isn't designed. It's derived."

AI broke two assumptions.

Assumption 1: Interfaces live on screens. - "Conversation has no viewport." An agent may have no interface at all until it needs to display something.

Assumption 2: Users initiate. - For 50 years, design assumed the user starts things. "Agents invert this. They act, then surface for approval. You respond." - The interface becomes "a negotiation you manage" rather than "a tool you operate." - A reactive interface cannot be a static layout — it needs context about surface, activity, urgency, and level of control needed.

Why hasn't anyone properly done this?

Because it's hard: state management across projections, context detection errors, uncanny shifting UI, exploding rule complexity. "These are real problems. Unsolved."

But the alternative — designing 45 interfaces per feature by hand — "is already impossible. We just pretend otherwise."

The potential paradigm shift

A new pipeline/model for intent-driven design: "Stop designing how things look. Start designing what things do."

Final line: "Intent is the interface now."

What if 'experience is just intent zones'.

Place intents (search, create, read, navigate) on a canvas. Hover any intent zone and see it rendered across desktop, mobile, voice, and API simultaneously. "One input. Infinite surfaces."

Author's Note

Written in one session using Claude Opus 4.5 through Quick Capture. Voice notes, discussion guides, and bullet points were dumped into the context window and shaped collaboratively through editing passes. The author notes: "I did make the frameworks manually though."