New SKUs from suppliers used to wait days for a copywriter. Now an AI pipeline turns raw spec sheets into on-brand titles, descriptions and spec tables, publishing draft listings the store team approves in one click — 12x faster to live.
Typical results
6 hrs
Spec sheet to published listing
160
New SKUs live per month
0
Backlog after one month
Figures reflect typical results for this kind of automation.
The challenge
Every shipment from suppliers brought dozens of new accessories, each arriving as a cryptic spec sheet in a different format. One employee rewrote them into product pages by hand — about 45 minutes per SKU — so new items sat unsold for three days or more, and a 900-SKU backlog kept growing.
Before and after
Spec sheet to live listing
−92%New SKUs published per month
4×Listing backlog
−100%The automation
We built a pipeline in Make that watches the supplier inbox, extracts every attribute from incoming spec sheets and files, and hands them to OpenAI with the brand's tone guide and category templates. The AI writes the title, description, bullet highlights and a formatted spec table, then creates a draft listing in the store complete with images and pricing rules. The team reviews a daily approval queue and publishes each listing in one click.
How it works
- 01
Spec sheet arrives
Supplier files land in the inbox and are parsed automatically, whatever the format.
- 02
Attributes extracted
Every spec, dimension and compatibility detail is mapped to the catalog schema.
- 03
AI writes the page
OpenAI drafts the title, description and spec table in the brand's own voice.
- 04
Draft in the store
A complete draft listing is created with images and pricing rules, ready to review.
- 05
One-click publish
The team approves from a daily queue and the page goes live instantly.
Starts from
- Supplier spec sheets
- New SKU emails
Delivers
- Draft listings
- One-click approval queue
- Live product pages
The impact
Time from spec sheet to live listing fell from three days to under six hours — 12x faster — and monthly new-SKU output quadrupled from about 40 to 160. The 900-SKU backlog cleared within a month, with every page written in a consistent brand voice.
Questions about this automation
What formats of supplier spec sheets can it read?
PDFs, spreadsheets, text documents and plain email text are all common inputs, including scanned sheets that need OCR. The pipeline extracts attributes such as dimensions, materials, compatibility and package contents and maps them to the store's catalog fields. Values it cannot read with confidence are marked for the reviewer instead of guessed.
How does the AI keep listings in the brand's voice?
The model works from a tone guide and a template per category, along with example listings the team already likes. It writes the title, description, bullet highlights and spec table in the same structure every time, and it only uses facts present in the spec sheet, so product claims stay accurate.
Does anything go live without a person checking it?
No. Every listing is created as a draft in the store with images and pricing rules applied, and it appears in a daily approval queue. A team member can edit it, approve it with one click or send it back. Simple automations like this are usually live within days.