The engine’s answer stops too early.
Chess engines give accurate, but opaque feedback. They are raw and hard to decipher, so it is hard to analyze a game when the engine does not say why.
LLMs also cannot reliably give an explanation, because they are optimized for text and only see notation, never truly seeing the board. That is what Decypherly fixes. By instilling true chess board understanding, the LLM can give accurate explanations in natural language conversation, something no other chess analysis product does.
Review that tells the story of the mistake.
Paste in a game. Decypherly finds the move where the position changed, explains the tactic, and shows the forcing line. The result feels closer to a coach than a scoreboard.
Why did I lose a piece here?
d5 looks safe — you attack it twice and Black defends once. But it’s a trap: after 6…Nxd5 7.Bxd8, the zwischenzug 7…Bb4+ arrives with check. You never get the queen back, and you’re left a knight down for a pawn.
Chess alone is a large market.
Chess is played regularly by hundreds of millions of people worldwide, and the majority of them are novices. They need a natural language engine they can ask questions to and get accurate explanations back, not raw, opaque engine lines.