Back to Commerce
Commerce Voice CardsAI Recommendations & Brand Voice

AI recommendations that sound like your brand wrote them

Commerce Voice Cards are auto-synthesised from your catalog on first connect. Peakhour infers your brand register, preferred vocabulary, and cultural context — no forms, no setup. Gets smarter with every campaign you approve or reject.

Auto-synthesised — no setup formsLearns from every approval and rejectionMarket-specific cultural vocabularyGets smarter after every campaign

The difference

Same product. Different voice.

Generic SaaS output

Dead Stock Clearance
Sale — Up to 40% Off
Clearance Items
Discounted Products

Peakhour in your brand voice

The Linen Edit — Last 12 Pieces
Season's End · Limited Stock
The Curated Clear-Out — 48 Hours
New Season In. Time to Move These Out.

Generated from your catalog. In your voice. Not a template — a learned synthesis.

How it works

From catalog to voice card to campaign copy

Four steps, fully automatic. You never fill a form. You never define a style guide. Peakhour reads your catalog and learns your voice from what's already there.

01

First catalog sync: Peakhour analyzes product titles, tags, and descriptions

When your Shopify catalog syncs for the first time, Peakhour's voice synthesis engine reads every product title, tag, and description. It's looking for vocabulary patterns, formality register, cultural markers, and the words you use most vs. the words generic SaaS systems use.

02

Voice card auto-synthesised: brand register, preferred vocabulary, cultural context

A Commerce Voice Card is created automatically. It captures: your brand's formality register, preferred category vocabulary, regional context, and a 'never use' list inferred from what's absent in your copy. No forms. No setup calls.

03

AI recommendations generated daily — titles in your voice, discount and duration suggested

Daily recommendations arrive with rail titles and campaign copy already in your brand voice. A fashion brand sees 'The Linen Edit — Last 12 Pieces'. A kitchenware brand sees 'The Summer Cookware Clear-Out — 48 Hours Only'. Same product type. Different voice.

04

Every approval/rejection teaches the voice card — gets smarter automatically

When you approve a title unchanged, Peakhour marks it as good. When you edit it (changing 'clearance' to 'curated edit'), it learns the substitution. When you reject, it learns the pattern to avoid. After 20–50 campaigns, the voice card is precise enough that you rarely need to edit.

Full feature set

Everything Commerce Voice Cards includes

Eight capabilities that turn generic AI output into brand-consistent commerce copy — synthesised from your catalog, refined by your behaviour.

Auto-synthesised on first connect

No forms to fill. On your first catalog sync, Peakhour analyzes your product titles, tags, and descriptions to build your initial voice card automatically.

Brand register detection

Peakhour infers your brand register: editorial or casual? Premium or accessible? Aspirational or practical? The voice card reflects how your catalog actually speaks.

Market-specific vocabulary

A fashion brand in New York uses different vocabulary than one in London or Sydney. Peakhour's voice synthesis is culturally aware — not just linguistically translated.

Category-specific terms

Different product categories in the same store can have different voice profiles. Footwear copy is different from home decor copy — Peakhour tracks both separately.

Learning loop from approvals

Every campaign approval, title edit, or rejection teaches the voice card. After 20 campaigns, Peakhour generates titles you wouldn't change. After 50, they sound like your copy team. It becomes your copywriter — one that gets better every week and never leaves.

Rail title generation

Every Smart Rail gets an AI-generated title in your voice — not a generic category label. 'The Linen Edit — Last 12 Pieces' instead of 'Dead Stock Clearance'.

Campaign copy in your voice

WhatsApp campaign messages, push notifications, and email subject lines — all generated in your brand voice, from your actual product data and intelligence scores.

Rejection signals refine the model

If you change 'Sale' to 'End of Season Edit', Peakhour learns that 'Sale' is not in your vocabulary. Negative signals are as valuable as positive ones for voice refinement.

The moat

“The Linen Edit — Last 12 Pieces” vs “Dead Stock Clearance”. That's the difference.

Generic commerce AI generates generic commerce copy. When every app uses the same underlying model, every Shopify store gets the same vocabulary — “sale”, “clearance”, “deal”. Commerce Voice Cards are trained on your store specifically.

A fashion brand in New York that uses “edit” and “pieces” gets those words back in every recommendation title. A kitchenware brand that never uses “limited time” never sees it. The moat deepens with every campaign — your competitor can't copy a voice card they don't have.

After 1st connectInitial voice card synthesised from catalog
After 10 campaignsVocabulary precision noticeably better
After 20 campaignsTitles rarely need editing
After 50 campaignsSounds like your copy team wrote it

Get started

Recommendations in your voice. Starting today.

Apply as a launch partner. Voice card synthesises automatically on first catalog sync. No setup, no forms, no style guide to write — just a better AI from day one.