StockMeta Assistant

An AI-assisted metadata workflow embedded in Adobe Stock Contributor for generating and applying titles and keywords.

Year: 2026
My role: Product definition · UX · Development · Launch
Platform: Chrome Extension

The problem

Creating isolated product images and patterns is only part of the stock workflow. Repeatedly writing accurate titles and ordered keywords for every asset becomes a slow, inconsistent bottleneck.

Product approach

I embedded the assistant where the work already happens. It reads the selected asset, calls a user-configured vision model, presents editable metadata and applies approved results to Adobe's form.

Key product decisions

In-context assistance

The tool lives beside Adobe's metadata form so users do not need to move images and text between products.

Human approval

Generated titles and keywords remain editable and are never submitted automatically.

Bring your own provider

Users configure their own compatible AI model, keeping the extension maintainable and costs transparent.

From need to launch

Observe

Mapped my own repeated workflow for isolated product images, patterns, titles and keywords.

Integrate

Connected image reading, vision-model output and Adobe form interaction inside one extension.

Safeguard

Added editable output, asset checks and explicit apply actions before publishing.

What I learned

AI features become products when they are embedded in a real workflow. Reliability, review controls and graceful fallbacks mattered as much as model output quality.

Store promo graphics

Screenshots from the Chrome Web Store listing.