Finds bestselling ebooks that hundreds of buyers rated three stars or less. Then it reads what they hated.
- Keyword
- EBOOK
- Runs
- When you ask
- Tested in
- Claude
- Last run
- 10 Nov 2026
Don’t trust me. Look it up.
Documented$1,995
For Iman Gadzhi’s program, Monetise. The pitch: AI writes the ebook, and you post it where buyers already hang out.
So how do we get that edge?
My adaptationHis program already is AI. The hard part is picking an ebook worth making. So find a bestseller people complain about, and fix what they hated.
You give itA niche, for example cookbooks.
- Pulls Amazon’s Kindle bestsellers for that niche.
- Drops what doesn’t belong, like magazines and memoirs.
- Checks each book’s 1 to 3 star ratings and its sales rank.
- Keeps the ones with hundreds of unhappy buyers that still sell.
- Reads the bad reviews and groups the complaints.
- Outlines a better book and drafts the posts. It never posts.
You getA table of gaps, what people hated, and an outline for the better book.
This is my real run.
Tested10 Nov 2026 · Amazon · Kindle cookbooks
72 cookbooks were left after cleaning Amazon’s top 100. Seven had hundreds of buyers at three stars or less, and all seven still sell.
| List rank | Ratings | 1 to 3 stars | People | Kindle rank |
|---|---|---|---|---|
| #51 | 19,962 | 9% | ~1,797 | #12,774 |
| #11 | 9,775 | 13% | ~1,271 | #2,540 |
| #39 | 7,824 | 16% | ~1,252 | #10,852 |
| #5 | 18,093 | 5% | ~905 | #882 |
| #13 | 5,993 | 12% | ~719 | #2,789 |
| #10 | 8,783 | 6% | ~527 | #2,441 |
| #100 | 3,411 | 11% | ~375 | #22,797 |
People = ratings × the share at 1 to 3 stars, rounded. Titles are hidden, like in the video. The prompt shows you the real ones.
No pictures. Way too many ingredients. Recipes that take forever.
Full reviews sit behind an Amazon sign-in, and I don’t sign in for runs. So the complaints come from Amazon’s public review topics.
Source: Amazon Kindle Best Sellers, Cookbooks, Food & Wine, 10 Nov 2026
What it can’t do.
- A rating isn’t a review. Three stars doesn’t say why.
- A gap is a lead, not a product. You still make the better book.
- My run stops at the list and the complaints. Selling the fixed book is untested.
- It never copies text or recipes.
- It drafts posts. You post them, by each community’s rules.
An AI that can browse the web. I tested it in Claude. If it’s a cookbook, you cook and test every recipe.
Now it’s your edge.
DaveOS
Or get the whole agent.
DaveOS connects to your AI and runs Gap Radar with you. You type one line. It does the job step by step and stops for your OK before anything gets posted.
Opening soon. $5 a month for the first 100 members.
DaveOS plugs into the AI you already use.
It opens on Claude first. The others follow as I test each one.