The map of AI jobs across an organization

XY Space
WF-R03Retail and ecommerce
T2 · Drafts for humansEngineered

Marketplace listing quality checks

Every night, live listings across the store and its marketplace channels get checked against the brand style guide and basic compliance rules — missing fields, banned claims, image sizing — and a fix list goes to the category manager instead of a shopper finding the problem first.

How it runs

ambiguousflagged
Trigger
Nightly listing scan runsShopify and marketplace channels
Extraction
Read the listing contenttitles, descriptions, images
Lookup
Check against the style guide and compliance rulesSharePoint
Rule
Flag listings below the quality bar
Branch
Hold ambiguous flags for review
Human gate
Category manager sign-off
Assemble report
Produce the fix listSharePoint
Outbound comms
Notify the listing teamMicrosoft 365 and Outlook
Outcome
Fix list issued
  • Trigger
  • Extraction
  • Lookup
  • Rule
  • Branch
  • Human gate
  • Assemble report
  • Outbound comms
  • Outcome

What it could return

Time a category manager would otherwise spend spot-checking listings for style and compliance issues by hand.

$10,120per year, returned
460 hrsper year, back

A rough estimate until this is measured on your own numbers. Every input above is yours to change.

Systems touched

Agentic skills run

Where the human sits

The workflow proposes a fix. Applying it is a category manager's call, made after they've looked at the flagged listing.

Confidence floor 70% — anything under that holds for a person instead of guessing.

Knowledge produced

This one doesn't produce any knowledge for memory to pick up.

Where it fits

Retail and ecommerce runs inside the commercial build. See the commercial build.

Where it sits

  1. 01The model
  2. 02Data
  3. 03Workflows
  4. 04Agentic skills
  5. 05Memory & knowledge
Workflows layer page