DEV Community

Tomi431
Tomi431

Posted on

Stop Manually Researching Products — Let Your AI Agent Do the First 80%

📸 Full version with screenshots: https://tomi431.github.io/Tomi/2026/08/04/ai-agent-product-research

A typical Amazon seller spends 3-4 hours on product research for every new product they consider. Open 8 browser tabs. Cross-reference prices. Copy numbers into a spreadsheet. Read competitor reviews. Check 1688 for sourcing costs. Repeat for 5 candidates. That is 15-20 hours — for information an AI agent can assemble in under 2 minutes.

The 80/20 rule applies here: 80% of product research is data gathering, which is pure repetition. 20% is judgment, which requires your experience. Automate the 80%. Focus on the 20%.

What the Agent Does in 90 Seconds

Give your AI agent a category name. It queries structured e-commerce data and returns:

  • HPI ranking: the top 10 products by Hidden Profit Index — composite scoring that surfaces products with high sales, low reviews, and strong organic traffic
  • Per-product detail: price, FBA fee, monthly sales volume, review count, star rating, brand, BSR rank
  • Trend data: 6-month sales trajectory, price stability, BSR rank improvement
  • Traffic analysis: what percentage of keywords are organic-only vs ad-dependent
  • Competition snapshot: brand concentration in the category, Amazon self-owned share 
# One command after installing sorftime-seller-agent:
# /goal Run HPI 5D verified shortlist for kitchen storage on Amazon US.
# stop after 70 turns
Enter fullscreen mode Exit fullscreen mode

What You Still Need to Do

The agent gives you a verified shortlist. It does not replace:

  • Your category expertise: you know which sub-niches have real margins and which are traps
  • Sample ordering: the numbers look good on screen — hold the product in your hands before committing
  • Supplier negotiation: 1688 prices are bait prices. Real quotes come from conversations
  • Competitive intuition: the agent tells you "this product has 75% organic traffic." You decide whether that traffic is sustainable

The 80/20 Split

Agent Does (80%) You Do (20%)
Pull and rank products Judge which ones fit your brand
Calculate margins Verify with real supplier quotes
Flag review patterns Read the actual reviews
Show price trends Decide if the trend will continue
Identify low-competition signals Decide if low competition means opportunity or dead market

The sellers who figure this out first will spend Friday afternoons reviewing shortlists their agent generated during the week. The sellers who do not will still be opening browser tabs.

Try it yourself: git clone https://github.com/DannylydST/sorftime-seller-agent → python3 scripts/install.py → get your key at open-intl.sorftime.com

Originally published with screenshots at https://tomi431.github.io/Tomi/2026/08/04/ai-agent-product-research.

Top comments (0)