OUR MISSION

Why GameMatch exists

You own hundreds of games and there is nothing to play. Storefronts show you what sells. We start from what you actually play.

The whole product is one action: build a pack

A pack is 2 to 10 games you loved, picked by you. That is all it is. From there we surface games that resemble them, one at a time, and you keep or drop each one.

Why go through that instead of deducing everything from your library? Because your library says what you bought, not what you feel like tonight. A pack is you setting the direction: three short, nervous games and three slow, narrative ones do not lead to the same place. You can keep several packs and switch depending on the mood.

The problem we care about

Stores recommend what performs. Top-10 lists recommend what everyone already owns. The same twenty games come back everywhere, and thousands of others stay invisible, including the ones you would have loved.

We start from the opposite end: your library, your real playtime, the games you kept coming back to. A recommendation is only worth something if it knows who it is talking to.

How the recommendations work

Every game in the catalogue is turned into a position in a taste space: a vector computed from its genres, its tags and its description. Two games that sit close together genuinely resemble each other, even when they share no keyword at all. We do not match labels, we match directions.

A pack is several of those positions at once. With a single game, we head in its direction. With two or three, we look for what they have in common: the flukes cancel out and your actual intent comes through. That is why a pack of three games you love beats any "games like X" list.

What changes over time

Your profile is not fixed. Every game you keep or turn down moves your taste vector, and the next recommendations take it into account. A game you turned down never comes back. Re-running a pack skips what it already showed you, so each run reaches a little further out.

The catalogue moves too. As the models improve, games are recomputed and the recommendations get better without you doing anything.

What we will not do

We will never show you a game that does not fit your taste, whatever the reason might be. It is the only thing that makes a recommendation worth reading.

Signing in goes through Steam itself: you approve the access on their side, and we read your library without ever writing to it.

Come and shape it

The project is young and it is built with the people who use it. If you have an idea, a game that was badly recommended, or a criticism, say it on Discord. That is where the decisions get made.