AZ Global Corp, e-commerce

An AI agent that finds the next winning product before the morning standup

AZ Global's buyers spent their mornings scrolling TikTok, Amazon best-seller lists and ad libraries to guess what would sell next, and by the time a product looked obvious, the market was crowded. Now an AI agent scans those sources every night, scores each candidate on demand, competition and margin, and hands the team a short list with evidence by 7am.

IndustryE-commerce

LocationUnited States

What we didAI agent and automation

Timeline6 weeks to launch, still tuning

Product scout shortlist: three products from TikTok, Amazon and Meta ads, each with a score, ready for review

The results

Measured by the client, over the period stated.

Dailyshortlist of scored products, ready before the team starts work

5sources scanned every night: TikTok, Amazon, AliExpress, Meta Ad Library and Google Trends

~3 hrsof manual research per buyer per day handed to the agent

The challenge

Product research was done by hand. Each buyer opened the same tabs every morning, copied promising items into a spreadsheet, then checked supplier prices and shipping times one by one. It depended on who was looking and what their feed happened to show that day.

Speed decides who wins a trend. A product that is rising on TikTok on Monday has dozens of copycat stores by Friday. AZ Global needed to spot demand early, drop the products with thin margins or saturated ads, and spend the team's time on deciding, not on searching.

What we built

An agent that runs every night. It collects trending and newly rising products, matches each one to suppliers to get real landed cost and delivery time, and checks how many advertisers are already running it. Every candidate gets a score for demand growth, competition and margin, and the agent writes a short reason for each one with links to the evidence. The top picks land in the team's shared sheet and chat at 7am.

The key decision was to keep a buyer in charge of the final call. The agent never launches a product on its own: a buyer approves or rejects each pick, and those decisions feed back into the scoring so the list gets sharper each week. Approved products are pushed to the store as drafts, with supplier, cost and suggested price filled in.