Idea Generator

The Problem with Ordinary Neural Networks

Ask any chatbot for ideas — and you'll get a safe, average, predictable list. The same one it gives everyone else. Ideas come out like stock photos: formally correct and belonging to no one.

What's Different Here

Under the hood — not one generator, but four independent ones that don't see each other's work. This very independence prevents ideas from collapsing into a uniform gray mass:

• Analogies — transfers working mechanisms from distant, unrelated fields: how coral reefs, medieval guilds, container shipping, airport dispatch work
• Inversion — breaks what everyone considers mandatory: “what if the opposite?”, “how to make it worse — and flip it”, “remove the main thing”
• 10x — only bold ideas, without regard for resources or permissions
• Combinatorics — breaks the task into parameters and crosses what no one has crossed

Then the volley comes together: duplicates and banalities are thrown out, survivors are rewritten in a bolder version, and you get ~20 numbered ideas — from grounded to wild, to feel the whole spectrum.

How to Use

Tell it what you need ideas for. A couple of clarifying questions — and the volley. Then you can:

• “20 more” — a new volley, deliberately away from what's already been covered
• “wilder” — only the boldest
• “expand #7” — one idea in detail: what it is, how it works, first step, five variations
• “cross #3 and #12” — a blend of two ideas
• “digest” — a summary of the whole session

You can also ask for a specific technique: SCAMPER, six thinking hats, premortem, question storming, “what would X do”.

What It's For

Business ideas and niches, content plans and formats, brand and product names, features, event scenarios, gifts, research topics, creative projects — any task where you need to go beyond the first obvious thought.

What It Doesn't Do

It doesn't evaluate, criticize, or explain why an idea is unrealistic. Generation is the whole point. You'll do the evaluating yourself.