Skip to content
UX Atlas

Practical workflows, with the prompts and the failure modes

AI for Design

Where a model genuinely saves a designer time, how to run the workflow, and what it still gets wrong.

10 entries across 8 categories

AI workflows10
Category
Level

10 entries

Using a vision model for a first-pass heuristic review, which is good at obvious defects and blind to context.

AI workflow · Evaluation

Turning a requirements document into a draft flow, which is most useful for the edge cases the document forgot.

AI workflow · Definition

Assisted coding of qualitative data, which speeds up the first pass and must not replace the analysis.

AI workflow · Research

Expanding a rough idea into a structured specification, which is mainly useful as a completeness check.

AI workflow · Definition

Using a model alongside automated tooling, which extends coverage a little and does not approach a real audit.

AI workflow · Evaluation

Generating and reviewing interface copy, which works well for volume and badly for voice.

AI workflow · Content

Drafting component documentation from code, which handles the mechanical parts and not the judgement.

AI workflow · Design systems

Use a model as a fast first reviewer against a named framework, then treat every point as a hypothesis rather than a finding.

AI workflow · Review

Models produce working interfaces quickly, and default to generic components with missing states. Supply the system, then review the parts they always get wrong.

AI workflow · Build

Everything in AI for Design, by category

The full contents of this section, so you can see the shape of it rather than only what a filter returned.