Everyone with a saving habit has the same problem, and it isn’t a shortage of ideas. It’s a pile. Bookmarks, a read-it-later queue, a notes app full of “there might be something here” — hundreds of inputs and no fast, honest way to look at any single one and answer the only question that matters: is there a business in this, or am I about to waste a weekend finding out there isn’t? Idea Harvester is the thing that answers it.
It’s a buy-once skill file — not an app, not a subscription, nothing to install. $12, one download, and it works inside the AI assistant you already use.
Who it’s for
If you clip everything and action almost none of it, this is pointed at you. Solo founders and indie hackers who don’t want to burn two hours researching every “is there a business in this” itch. Creators sitting on a graveyard of saved threads, teardowns, and “I could automate that” essays. You don’t have to be technical — if you can paste text into Claude, ChatGPT, or Gemini, you can run it. The models are already capable of the judgment; what’s missing is a framework that makes them deliver a verdict instead of a summary.
What it actually is
Idea Harvester is a single skill file. You paste it once into your assistant of choice, and from then on you hand it a link, a PDF, or pasted text and it runs that input through a real assessment instead of politely describing it back to you:
- Review it — a 30-second worth-it-or-skip call on a link, nothing heavier.
- Harvest it — the full breakdown of an article: the core thesis, who would genuinely pay, whether one person could build it, roughly how long to a first dollar, and who already competes.
- What can I sell here — specific revenue paths from a report or note, ranked by how fast each gets you to cash.
Underneath all three modes, every input passes the same five questions before anything reaches you: Who specifically needs this? Could one person build it? How long to the first dollar? What already competes, and where are the gaps? Is there an angle nobody’s working yet?
The outcome you’re buying
A position, not a wall of text. Every answer lands on one of three words — do this, look into, or skip — with the reasoning attached. “Build the niche calculator, live in about a week, here’s the angle the current players keep missing.” Or: “Skip the directory — too much curation for too little early money, but the automation underneath it is the actual product.” No “it depends.” You can act on it or argue with it, and both beat the pile.
The part that matters most is the gate at the end: is this real, is it tested, would you actually put your name on it? If the answer is no, it doesn’t dress the idea up and hand it over anyway. It’s the difference between a tool that calls every idea brilliant and one that will occasionally look at your pitch and tell you to go do something else. You get structured judgment and estimates — where the opportunity probably sits and whether it’s worth your next free Saturday — not guarantees or a promised revenue number.
Why twelve dollars and not free
There are free prompts that promise roughly this. Most hand you a confident summary with no framework underneath. Idea Harvester was built and run against real articles before it went up for sale: three detection modes, a proper assessment matrix, research folded in, and that quality gate on the end. It’s made for one specific buyer instead of trying to be everything for everyone. Twelve dollars, paid once — a read-it-later app costs more than that every month and never makes a single call for you. You get the skill file, a full worked example so you know exactly what you’re buying, a two-minute quick-start, and every v1.x update. Full refund inside 30 days.
You own the file outright. No signup, no server, no subscription.
Idea Harvester — $12, one-time. Instant download. Buy it on Gumroad →
The build-log — how this came out of my own bookmarks graveyard — lives over here.
Hugh Fletcher
Founder & builder at Atomic24.