Seller input prototype In progress
A front-end prototype that compares a seller's new settings with similar sellers and routes large changes to a person.
Recreated on dummy data. Every name, figure and date in the recreation is invented.
Result: In progress. I built a similar system at Amazon and am close to putting this one in place.
Problem: Merchants and partners change settings that move delivery promises, such as a 48-hour pick-pack-prep time. Someone has to judge each change, and reviewing every one by hand does not scale.
What I built:
- A step that places the seller among similar sellers.
- A comparison of the change’s expected impact against that group’s baseline.
- Three outcomes: accept a trivial change, monitor and review a non-trivial one, and send any change above a gross merchandise value (GMV) threshold to a person right away.
How I worked with AI: I built this front-end prototype with Claude Code in VS Code, from my own description, on dummy data.
Decisions and trade-offs:
- People review only the changes that matter, and the rest go through on their own. The cost is that a trivial change can go live unchecked.
What changed: Not yet measured.
Links: The prototype is below. Dummy data generator ยท Dummy data