All projects

Case study, 2026

Shopify Automation

An anonymized body of Shopify work for a niche e-commerce retailer, beginning with a distributor-catalog pipeline the owner can operate without me.

Role
Founder & Engineer, Cannon AI LLC
Status
Client delivery

Anonymized case study · client system stays private

Problem

A small retailer was manually rebuilding distributor listings in Shopify, making recurring catalog entry slow and error-prone.

Solution

I built a scraper that pulls the distributor catalog, maps it to Shopify's product schema, and creates reviewable drafts through a hosted control panel. That first delivery became the foundation for an in-progress storefront workstream and a separately scoped operations backlog.

Impact

The catalog system was delivered and paid on June 11, 2026. The client can run it without me. Storefront work is in progress; the operations backlog has not started.

Public work ledger

One engagement, shown at three honest stages.

Delivered work is separated from private work in progress and ideas that have only been scoped. Visuals are synthetic; client systems stay private.

DeliveredJune 11, 2026

A reviewable catalog system the owner can run.

The delivered workflow pulls source catalog data, maps it into product fields, and creates drafts for human review. Nothing is published by the automation.

Delivered and paid
Client-operable handoff

Synthetic flow

Drafts before publish

01Source catalogPull and clean
02Product mappingNormalize fields
03Draft queueHuman review
Automation preparesHuman publishes

Build notes

  • Python
  • Shopify API
  • FastAPI
  • Railway
  • BeautifulSoup

The first problem

A small e-commerce retailer was rebuilding distributor listings in Shopify by hand, one product at a time. The repeated entry was slow and made catalog mistakes more likely.

The catalog system

  • A scraper pulls the distributor's catalog and cleans it up into a consistent format.
  • A mapping layer translates that into Shopify's product structure — titles, variants, images, pricing, metafields.
  • Products land as drafts, so nothing goes live without a human review.
  • A hosted control panel lets the owner run imports herself, watch progress, and re-run on demand — she keeps using it after I'm gone.

Why the handoff matters

  • Built to be handed off, not hovered over: the owner operates it without me.
  • Drafts-first by design — the system never publishes to a live store without review.
  • Covered by a deep test suite, because a catalog tool that silently corrupts product data is worse than no tool.
  • Led with a working build and a reviewable draft workflow rather than a promise about future automation.

A broader Shopify practice

The catalog pipeline is delivered. Follow-on storefront and operations work is tracked separately and presented here by stage so scoped ideas are not mistaken for shipped systems.

The client stays anonymous by design. Public visuals use synthetic product names and prices; they do not expose the store, supplier, credentials, customer data, private screenshots, or live admin interfaces.

Why It Matters

This is the whole consulting thesis in one relationship: find a real cost a business is paying every day, build the smallest reviewable system that removes it, and hand over a tool they can own. Continued work is stronger evidence than a polished sales claim.