All brand systems Part of the AppOmni collection

AI-Assisted Design Workflows

AI-assisted workflows for turning design intent into reusable web components, marketing UI, and motion.

Company
AppOmni
Role
Senior Visual Designer
Three selected frames from the original Marlin AI motion piece -- "Surface," "Prioritize," and "Remediate" -- each showing a staged interface reveal

The work

At AppOmni, I developed workflows that used AI to help translate designed sections, product references, and motion principles into reusable production assets. I set the design intent, structure, references, and acceptance criteria, then directed analysis, implementation, variation, and correction.

Context

AppOmni's marketing and product-storytelling pages were growing more complex, and needed to move faster without drifting from the established brand and product systems.

Challenge

Designed sections, product references, and motion principles needed to become reusable production assets instead of one-off compositions rebuilt for every page.

Contribution

As Senior Visual Designer, I defined the design intent, component structure, references, and acceptance criteria for each workflow, then directed the AI-assisted analysis, construction, and correction.

Approach

I reused existing components and library styles wherever possible, added new patterns only when needed, and supplied the tools with brand and product references. The work used configured AI tools, not a trained proprietary model.

Outcome

Page production time dropped from 10 days to 3 days after the reusable component system was introduced. The result came from the full production system, not AI alone.

10 days to 3 days

Page production time after introducing the reusable component system

AppOmni

The production problem

Complex marketing pages needed to move faster without drifting from AppOmni's established brand and product systems. The goal was not to make AI the outcome. It was to build a more reusable and consistent production system, with AI assisting inside that system rather than replacing the judgment that shaped it.

Component workflow

From composition to component

I began with a page section I had designed, then defined its reusable structure: what should stay fixed, what should vary, and which existing components and library styles could be reused instead of rebuilt. AI assisted with analyzing that structure and implementing the first pass; I evaluated the result against the original composition and directed the corrections.

The interface below is a recreated demonstration of this workflow, built to illustrate the process. It is not the AppOmni production environment, and the inspector-style properties panel shown did not exist natively inside AppOmni's own Figma files.

Recreated "Component Workflow / Studio" interface showing a published AppOmni section reconstructed into an editable component, with a properties panel, a proposed component structure, and a generated component specification
A recreated design-tool inspection view of the same AppOmni feature section, showing its layer structure, nested components, and exposed properties (viewport, theme, layout, card count) in a Design/Prototype panel
Recreated demonstration: illustrates how the section's structure and properties could be exposed for reuse, not a screenshot of AppOmni's own Figma file.

Variants

Expanding the system

Once the base component existed, the workflow extended it across the combinations a real page actually needs: desktop and mobile, light and dark, different content lengths, and different icon, label, headline, copy, and CTA combinations. Unless noted otherwise, the variants shown here are reconstructed examples built to demonstrate the approach, not screenshots pulled directly from AppOmni's own files.

Recreated "Create variants" step showing the same AppOmni feature section recomposed across desktop and mobile, dark and light theme combinations, with icon, label, and content variations
A component-set view in the recreated workflow grouping four variants of the same AppOmni feature section by viewport (desktop/mobile) and theme (dark/light)
Recreated demonstration of how the variant set could be organized for reuse.

Motion

Motion as a system

Motion supported the storytelling itself: progressive headline reveals, staged UI builds, a sense of depth and forward movement, deliberate attention shifts, and technical storytelling that needed to hold up on its own, without narration.

The Marlin AI animation below is original work. The motion-system interface shown afterward is a recreated demonstration inspired by that work. It is not the production editor used at AppOmni, and the original timing tokens, easing curves, and full accessibility specifications were not available.

Marlin AI

I defined and reviewed

  • Communication goals
  • Component logic
  • Brand and product references
  • Information hierarchy
  • Acceptance criteria
  • Corrections and final design judgment

AI assisted with

  • Layout analysis
  • Initial construction
  • Variant generation
  • Implementation
  • Debugging
  • Repetitive production tasks
  • Export preparation

Outcome

What changed

Page production time dropped from 10 days to 3 days after this reusable component system was introduced. That result reflects the full system: a reusable setup, more consistent references, adjustable rules, less repetitive construction, and a clearer handoff from design to development.

Creative Systems Toolkit

Supporting tools

Alongside the production workflows above, I built a set of small, focused utilities to test and extend the same design system: a button-state lab with per-state color-contrast ratios, a generative dot-pattern tool, a brand-logo placement composer, and a WebP export demo. These combine observed design-system references with clearly labeled recreated demonstrations and proposed extensions.