Core Workflows
End-to-end flows for creating storefront pages
Core Workflows
1. Ad Creative to Landing Page
Turn a Meta/Google ad into a persona-matched, published landing page.
list_workspaces()
→ get_brand_kit(workspace_id)
→ upload_ad_creative(file_url, workspace_id)
→ analyze_ad_creative(image_urls)
→ match_persona_to_ad(ad_analysis, workspace_id)
→ list_industry_templates(industry)
→ generate_pcx_page(template_id, ad_analysis, persona_match, workspace_id)
→ create_page(blueprint_id, store_id, workspace_id)
→ publish_page(page_id, workspace_id)
→ public URL ready for ad platformKey Rules
- Hero headline must match ad headline exactly (message match improves Quality Score)
- All copy uses persona vocabulary from
copy_adaptations - Never invent product details, only use what's in
ad_analysis - Slug format:
{brand}-{product}-{persona}-v{n} - Always
validate_blueprintbeforecreate_page
2. Brand Brief to Live Page
Generate a storefront page from a brand brief.
list_workspaces()
→ get_brand_kit(workspace_id)
→ list_design_sources(workspace_id)
→ list_component_types()
→ generate_blueprint_prompt(brief, page_type)
→ [AI assembles blueprint JSON]
→ validate_blueprint(json)
→ write_blueprint(slug, json)
→ screenshot_blueprint(slug, "both")
→ create_page(blueprint_id, store_id, workspace_id)
→ publish_page(page_id, workspace_id)Quality Checklist
- Theme uses brand kit colors (not defaults)
- Hero heading uses
theme.typographyfor weight/tracking - Stats section has bold accent color or contrasting background
lx_buy_box variant="sticky_bar"present on every PDP- All icons use valid
lucide:nameformat validate_blueprintpasses with zero errors
3. A/B Testing
Create experiments, measure results, scale winners.
create_page(blueprint_a, store_id, workspace_id) → page_id_a
duplicate_page(page_id_a) → page_id_b
[modify blueprint B — different hero, CTA, etc.]
create_ab_test({
page_ids: [page_id_a, page_id_b],
weights: [50, 50],
workspace_id
})
# Wait for traffic...
get_experiment_results(experiment_id)
→ cvr_per_variant, significance, winner_recommendation
scale_winner(experiment_id, winning_page_id)
→ 100% traffic to winner, experiment complete4. Brand Setup
Load brand identity and capture design references.
list_workspaces() → workspace_id
get_brand_kit(workspace_id)
update_brand_kit(workspace_id, {
primary_color, font_heading, font_body, logo_url, voice_md
})
capture_design_source(url, workspace_id)
list_design_sources(workspace_id)Use brand kit data when generating any blueprint to ensure consistent identity across all pages.