How Mira Works
How Mira turns a chess position into an explanation, why that can be difficult, and how your feedback helps us make her better.
First, Mira looks at the position
Mira uses a chess engine called Stockfish to examine the board and compare possible moves. You do not need to know how Stockfish works: think of it as the part that does the calculation before Mira tries to explain the result.
In this public build, Mira can give hints and explain the previous move. Follow-up questions about the position are not connected yet, so the chat box is currently a preview of a feature still being built.
The harder problem: explaining why
Finding a strong move and explaining it to a particular player are different tasks. In a complex position, Mira may focus on the wrong detail, list several moves instead of answering the question, explain what happened without explaining why it matters, or sound more mechanical than helpful.
The difficult part is working out what the player is trying to understand and explaining the most useful idea in a clear, concise way — like a coach sitting beside the board.
How the pieces fit together
Why your feedback matters
A rating tells us that an answer missed. A short note tells us how. When you say what you expected to hear, we can see whether Mira got the chess wrong, explained it badly, or simply answered a different question from the one you meant.
Less useful: “Not helpful.”
More useful: “I wanted to know why this move was dangerous, but the answer only listed variations.”
You do not need technical language. Just describe the gap in your own words: what were you trying to understand, and what did Mira explain instead?
Your feedback does not retrain Mira immediately. We review it to find recurring problems, reproduce weak answers, and decide what to improve and test next.
What is still ahead
Full opening courses, Puzzle Mode, long-term personalised learning memory, and broader language support are planned for future updates. The goal is not only stronger analysis, but explanations that are clearer, more relevant, and more useful to real players.