Sizing an Idea Down to a Weekend

Find the One Thing

Every idea arrives as a system: accounts, payments, notifications, an admin panel, a mobile version. That version is a year of work, and you will abandon it in week three. The move is to find the single action that makes the idea worth anything at all, and build only that. A marketplace is really "someone can list a thing and someone else can see it". A scheduling app is really "a client picks a slot and I get told". Everything else is support scaffolding around that one action. Strip it back deliberately, on paper, before you describe anything to an AI. If you do not choose the scope, the AI will cheerfully attempt all of it. What comes back will be broad, shallow and impossible to fix.

• Write your idea as a single sentence containing exactly one verb
• Ask of every feature: does the core action work without this? Then cut it
• Accounts, admin panels and settings are almost never version one
• If you do not set the scope, the AI will attempt everything at once

Fake Everything You Can

A first version does not have to do the work. It has to prove the work is worth doing. Notifications can be an email to you that you forward by hand. A payment can be an invoice you send yourself. Matching, moderation, approval, curation: all can be you, quietly, at a laptop, for the first twenty users. This is not cheating. It is the oldest trick in product development, and it has a real advantage. Doing the job by hand teaches you exactly what the automated version needs to do, and you cannot get that knowledge by guessing. Automate a step only after you have done it by hand enough times to be bored by it. Boredom is a reliable signal that you now understand the rules well enough to write them down.

• Doing it by hand first teaches you what to automate and what to skip
• Manual steps are invisible to users when volume is low
• Automate the step you are bored of, not the step you are excited about
• Every faked feature is a feature you did not have to debug this weekend

A Weekend-Sized Test

Here is a check before you start. Can you describe what you are building in three sentences? Can you name the one person who will use it first? Can you say what you will learn from it? If any of those is fuzzy, the project is too big or too vague. You will lose the weekend to indecision rather than to the tools. It also helps to decide in advance what "done" means for this version. There is no natural stopping point — you can always add one more thing, and the AI will happily add it. So write your finish line down before you start. A line you have already written is much harder to move than one you are holding loosely in your head.

• Three sentences, one first user, one thing you will learn
• Write the finish line down before you start building
• Vagueness costs more weekend hours than difficulty does
• If you cannot describe it, you cannot prompt for it either

Try It Yourself

Shrinking an idea is a skill you only learn on your own idea. Take the one you keep coming back to and put it through the four cuts, on paper, before you open any tool.

My full idea: [the whole system, exactly as you picture it — every feature]

The one action that matters: [one sentence with exactly one verb]

What I could fake by hand: [the features you would quietly do yourself for the first twenty users]

The weekend version: [what it is, who tries it first, what you will learn from it]