Chat gets bolted onto tasks with a known shape, adding turn-taking, typing, and interpretation variance to something a control could do in one gesture.
← Back to patternsGoogle
Inline & Contextual
Chat Off-Ramp
When the best AI interface is no conversation at all
Chat gets bolted onto tasks that have a known shape — and suddenly cropping an image takes three turns of typing. When the task's parameters are enumerable, the strongest move is the one this library keeps circling: skip the conversation entirely. Ship the task-native control — a toolbar, a slider, a button — with the model behind it, and keep a free-form escape hatch for the long tail. NNGroup's most radical finding, distilled: sometimes the fix is not a better surface next to chat, but no chat at all.
Metaphor
Off-Ramp
Interaction
Task-Native Controls, One-Gesture Actions, Escape Hatch to Free-Form
Use case
Photo Editing, Repetitive Structured Tasks, Known-Shape Workflows
Failure mode
Chat as Overhead
Control mechanisms
Direct Controls With AI Behind Them · Enumerated Parameters · Optional Free-Form Fallback
Reference examples
Adobe Photoshop Generative Fill toolbar · Google Photos Magic Editor · Canva Magic Studio tools
Interactive Example
The live side-by-side is the comparison.
This prototype already contains the contrast between chat and the pattern response. It stands as the primary comparison surface for this page.
A crop is a rectangle, not a topic for discussion.
Six scripted turns, 38 typed words, two mock minutes — and the first square crop still clipped the lighthouse out of frame. Two of the six turns were the model asking for a value a control could have offered: which part, how much warmer. Every clarifying question marks a control that was never built.
The same model sits behind four crop presets, two 0–100 dials, and two one-tap suggestions — including a straighten button that names the tilt it found, −2.2°. The model still contributes judgment; it just stops negotiating for it. Three gestures, nine seconds, and the horizon is level.
The hatch matters as much as the toolbar: a free-form field under it, labelled unused so far, catching the intents no preset can enumerate. Enumerable parameters get controls, the long tail gets language — Photoshop's Generative Fill and Google Photos' Magic Editor both land there, an AI capability shaped as a tool with prose kept in reserve.
When every parameter of a task can be named, the conversation is not the interface — it is the tax.
DOPE Lens
How this pattern improves AI interaction quality.
Observability
The available actions are the interface — every capability is a visible control, none hides behind phrasing
Directability
One gesture does what three chat turns approximated, and the free-form hatch stays open for the unenumerated case
Predictability
A labeled control does one named thing — no turn-taking, no interpretation variance between runs
Explainability
The control's label is the explanation — what it does is what it says, before it runs
Fit
Where it earns its place.
Use it
when
when
The task's parameters are enumerable and repeated — a conversation adds turns where a control would do.
Avoid
when
when
Open-ended, exploratory, or genuinely conversational tasks — when intent cannot be enumerated, chat earns its place.
In the wild
Where this pattern already ships.
Real implementations from products users already know. Each entry shows how the pattern got applied — what shipped, where it sits in the product, what it traded off.
Screenshot pending
Adobe Photoshop · Generative Fill toolbar
Adobe
Generative AI behind a selection tool and a button — the model does the work, the interface stays a toolbar. The optional prompt field is the escape hatch, not the primary surface.
Visit Adobe ↗
Screenshot pending
Google Photos · Magic Editor
Tap, drag, erase — generative editing with zero conversation. Suggested edits appear as buttons; parameters the task actually has (position, size, presence) map to gestures.
Visit Google ↗
Screenshot pending
Canva · Magic Studio tools
Canva
A suite of single-purpose AI tools (Magic Eraser, Magic Expand, Magic Grab) shipped as named buttons in the editor — each does one enumerable thing, no dialogue required.
Visit Canva ↗
Research Frame
Why this pattern exists, and where it breaks.
When the task's parameters are enumerable, skip the conversation: ship the task-native control with the model behind it, and keep a free-form escape hatch for the long tail.
This is the meta-case — the conversation itself is the overhead; there is nothing for the dialogue to disambiguate.
Fixed controls cap the ceiling — without an escape hatch to free-form input, the unenumerated 20% of intents has nowhere to go.
Common questions
What people ask about this pattern.
When should I use the Chat Off-Ramp pattern?
+
The task's parameters are enumerable and repeated — a conversation adds turns where a control would do.
When should I avoid the Chat Off-Ramp pattern?
+
Open-ended, exploratory, or genuinely conversational tasks — when intent cannot be enumerated, chat earns its place.
What problem does Chat Off-Ramp solve?
+
Chat gets bolted onto tasks with a known shape, adding turn-taking, typing, and interpretation variance to something a control could do in one gesture.
Why is chat the wrong fit for this?
+
This is the meta-case — the conversation itself is the overhead; there is nothing for the dialogue to disambiguate.
Related Patterns
Curated connections, not just neighbors.
Hand-picked relationships — what this pattern pairs with in practice, what it replaces, what it builds on, and what builds on it.
Alternative to
Disambiguation Branch
Disambiguation Branch lets you pick a reading of an ambiguous prompt; Chat Off-Ramp's labeled controls leave nothing to misread.
Extends
Property Panel
Property Panel puts controls beside the prompt; Chat Off-Ramp drops the conversation once every parameter is a control.
Extended by
Select & Transform
Select & Transform is the off-ramp for text edits: select the span and pick a verb from the toolbar instead of describing it.