Built by Fahad Cheema
How two n8n workflows took an online store’s entire inventory operation off human hands.
The client is a ready-to-wear ecommerce store based in Canada
The whole operation ran out of one Airtable base and a Shopify storefront. Airtable held the truth: every product, every size, every stock count. Shopify was where customers actually bought.
The catalog had grown past 600 products and kept growing. The systems that worked at 30 products did not work at 600.
Two problems, and they fed each other.
The photos were not store photos. Every product had a couple of phone snaps taken to track stock. Bad lighting, cluttered backgrounds. Useless for a storefront, so products sat unlisted.
There was no real sync. Every product that went live was typed into Shopify by hand, one at a time. Variants, inventory, images, metafields, categories: all manual. The two systems drifted apart the moment anyone touched either one.
The goal was to take the human out of the middle. No photographer, no one-by-one uploads, no spreadsheet kept in sync by hand. The system was built as two n8n workflows, designed around three core components.
n8n is the central hub. It pulls only unprocessed records from Airtable so it never repeats work, and runs in batches with a short wait between calls so no API ever throttles it. Every step is idempotent, so the same workflow can run again and again and the result stays clean.
A code node builds a prompt from the real garment data: brand, fabric, design type, color and cut. It even detects one-piece items and dupatta presence. That prompt and the phone photos go to nano-banana-2, which returns a clean, well-lit studio image. The URL is written straight back to Airtable.
Core product and inventory operations run on the REST Admin API. Categories, metafields, metaobjects and media run on the GraphQL Admin API, where Shopify exposes them properly. Airtable stays the single source of truth. Shopify reflects it on every run.
How the two workflows run
The image workflow runs: Airtable → n8n → fal.ai → Image URL → Airtable.
The sync workflow makes one decision per product: has this been created in Shopify before?
The photography line item, gone
Professional apparel photography runs $50 to $75 per image, and each product needs a front and a back shot. That is roughly $100 to $150 per product. Across 600+ products, a full catalog shoot is $60,000 or more, and that is before the two to four weeks of studio, styling and retouching time it takes to get through that many SKUs.
The fal.ai workflow does the same job at cents per image. The entire catalog re-shoots itself for a few hundred dollars in API cost, in an afternoon, with no studio to book and nothing to coordinate.
The manual listing work, gone
Typing 600 products into Shopify by hand, with variants, inventory, metafields, categories and images, takes a careful 10 to 15 minutes each. That is 100 to 150 hours of data entry, the better part of a month of full-time work, repeated every time stock or prices change.
The sync workflow does all of it in a single unattended run. Variants, inventory, metafields, categories and images move together every time, and because each step is idempotent it is safe to re-run and safe to scale.
The bottom line: a one-time setup replaced a recurring five-figure photography bill and a month of manual listing work, and turned both into a workflow that runs on its own. Adding a new behavior later (low-stock alerts, new-arrival posts, pricing rules) is a new branch, not a new project.