Case study 02
Agents · Browser automation
Dynamic Shopify Storefronts
Someone clicks an ad and lands on the same product page as everyone else. This Shopify feature changes that. It pulls the ad's creative and audience from Meta, captures the product page with Playwright, has an AI agent suggest edits that match the ad, and publishes the version the merchant approves.
My role
I built it end to end, from the Meta Ads integration and creative analysis to the page agent, preview, publishing and A/B analytics.
Core technologies
System path
A simplified view of the product workflow.
- 01
Meta ad context
- 02
Shopify context
- 03
Browser capture
- 04
Page agent
- 05
Merchant preview
- 06
Publish + attribute
01
The idea
Landing pages that match the ad
If an ad makes a specific promise to a specific audience, the page should pick up where the ad left off. We keep the ad's context and use it to generate a variant of the product page for that ad.
02
My work
Capture, edit, review
- 01Pull campaign, audience, creative, copy and performance data from Meta Ads.
- 02Combine it with the Shopify product, page and brand data, plus whatever the merchant asks for.
- 03Capture the mobile page (visible DOM, text and screenshots) in a real browser.
- 04Let the agent plan and apply copy, media and layout changes.
- 05Show the merchant a side-by-side view and a live iframe preview before anything goes out.
03
Publishing
Not breaking live stores
Editing someone's live storefront is risky, so every change goes through selector checks, visual verification, saved states and health checks. Approved variants are published through Shopify app extensions and metafields instead of rewriting the theme.
Each visitor gets assigned a variant once and keeps it. Shopify Web Pixels report page views, add-to-carts, checkouts and purchases back to us, so merchants can see which variant actually sold more.