Asking in the Shape It Understands
Why a natural-language question fails and a slightly different one works.
What it is for
It is a computation engine wearing a search box. A natural question fails and a slightly different phrasing succeeds for a reason worth understanding once — this covers the shape it actually parses.
The skill
It is a computation engine with a natural-language front end, not a chatbot. The front end fails in predictable ways. What works: - **Units in the query, always.** `50 km/h to mph`, `3 tablespoons in ml`. The unit is what selects the right computation. - **The operation named**, not implied. `integrate x^2 sin x dx` beats "what is the integral of...". `solve x^2 - 5x + 6 = 0` beats "find x". - **One question.** Two clauses joined by "and" are parsed as one query and usually as neither. - **Explicit variables** in anything symbolic: `d/dx` rather than "derivative". What fails, and the fix: - A question with a pronoun or a reference to the previous query — there is no conversation state. Repeat the subject. - A word with a units meaning and an everyday meaning. "second", "minute", "mole", "pound". Disambiguate: `pound (mass)`. - Anything that needs a source rather than a computation. It will answer, and the answer's currency is whatever its data set says. Always open **Step-by-step solution** when there is one. It shows the method, which is what tells you whether it understood the question you meant.
How to install it
1. Open the tool and start a new conversation, project or template. 2. Paste the text below wherever the tool keeps standing instructions — usually called system prompt, instructions, or persona. 3. Save it so it applies to every run rather than a single message.