Roast My Idea: the full run on an example input

What it read

Idea stage, nothing built, solo at 15 hours a week, bootstrapped, the only encouragement so far from Discord friends.

The read

You are at the cheapest moment this idea will ever have: nothing built, a clear concept, and the instinct to check before sinking your 15 hours a week into it. That instinct is rare and it is right. The fear under the question is that Discord friends saying cool is the only signal you have, and you can already feel it is not a buyer. The verdict is Sharpen it: the idea lives or dies on whether a developer trusts a prediction from a description more than the playtest they can run for almost nothing.

Medium confidence

The idea and the buyer are clear enough to find the crux, but there is no price, no named type of developer, and no conversation with a stranger yet, so this is medium, not high.

The room

Friend. I would paste my next concept in once out of curiosity. I cannot see a reason I would come back a second time unless the first answer surprised me and turned out right.

Customer. I would pay for an answer I can act on before I commit months. I would not pay for a score I cannot check until the game is built, because by then my own playtest tells me the same thing for free.

Cofounder. Before giving this my evenings I would need one case where the prediction disagreed with a developer's gut and the developer changed what they built because of it. Without that, it is a toy that sounds smart.

Investor. What I would attack is the proof of accuracy: you have no outcome data to show the predictions are right, and that data lives with the studios and platforms, not with you. That makes this a side project until someone pays for it.

What's really going on

The real question. Trust in the answer, not reach. This is a product problem before it is a distribution problem. Reaching game developers is doable; they gather in Discords and forums you already sit in. The hard part is that your answer is a prediction they cannot verify until after they build, while a playtest gives them real players and real behavior. Until a developer believes your number over their own small test, finding more of them only finds more people who say interesting.

The competitor you missed. The free playtest is your real rival. The incumbent is not another prediction tool. It is Steam Playtest, itch.io demos, a Discord playtest channel, and paid services like PlaytestCloud, plus any general chat assistant a developer can paste a description into for free. Each one either gives real player behavior or a quick opinion at no cost. Your tool has to beat both: more trusted than a chat assistant's guess, and earlier than any playtest.

This is your wedge. Sell the choice before anything is playable. The only window where a prediction beats a playtest is before a prototype exists, when there is nothing to test. Position it as: pick which of your three concepts to prototype first, and get the archetype read in a minute instead of a month. The trigger moment is a developer stuck between ideas after finishing a project or before a game jam. As described now it reads like a feature inside a design or pitch tool, and it only becomes a product of its own if developers return to it for every new concept.

The fatal flaw

Developers trust a cheap playtest or demo retention data more than a prediction made from a description. A prediction from text cannot be checked until the game exists, and once the game exists the developer has real players to watch for almost nothing. That squeezes your value into the short window before a prototype, the moment a developer is most attached to the idea and least willing to pay to hear it is weak. If no developer will pay for that early answer, the idea has no buyer, however good the archetype model gets. Naming it now costs you a few messages; finding it after months of building costs you the months.

Your prescription

List 10 game developers you know by name, Discord friends included, and write down one price per prediction before you talk to anyone. By 2026-11-27, show each of them the tool idea at that price and ask for a straight yes or no. If 2 of the 10 say yes to paying it, you have a buyer worth building for. If only your friends say yes, or nobody does, the trust problem is real and the next step is the concept-choice angle, not the build.

Good sign: you're onto it. A developer asks to run a concept they are actually stuck on right now, or asks how soon they can pay, before you push. That moment proves the answer has value in their real decision, not just curiosity.

Warning: rethink the format. The replies feel warm and vague: sounds cool, would love to try it, let me know when it is out. That feeling means they like the idea of the tool and do not need it, and a price will not survive it.

The full read

How it works. Developers pay for answers that change a decision. Your answer arrives before the game exists and cannot be checked until after, so it competes on trust alone, while a playtest competes on evidence. To earn trust you would need proof that past predictions matched real player behavior, and that outcome data sits with studios and platforms, not with a solo builder. No trust, no willingness to pay, no matter how clever the archetype logic is.

How it plays out. The usual curve: a couple of months of evenings building the model, a launch to the same Discord, a wave of friends running one concept each for free. They say interesting. Nobody runs a second one, because nothing they built proved the first answer right. By month three or four the tool is accurate in theory and unused in practice, and the question becomes whether to add more archetypes.

The trap to avoid. The trap is treating Discord encouragement as demand, then building the model first on the belief that accuracy will create buyers. Friends saying yes is emotionally true and commercially meaningless, and a better model does not fix a buyer who does not trust predictions. Ask for the price before you write the code, because a no now is cheap and a no after the build is not.