💭 in your head · nothing to type
Picture two conductors before a concert. One receives a full score — every note, tempo marking, dynamic, rest. The other hears only: "Play something beautiful."
Run it in your head: both orchestras take the stage. What happens next?
Now catch the parallel: was your last AI prompt a score — or a "something beautiful"?
💭 in your head · nothing to type
Hold the image of an architect at a drafting table.
On the desk: either a technical plan — dimensions, materials, load tolerances — or a note that says "make it look nice." Picture what each one actually enables the architect to build.
Notice where in your prompts you had a blueprint, and where you had a wish. Which one steers the outcome — and which one hands it to chance?
Context = who you are and what's the situation
Task = what exactly needs to be done
Instructions = how, in what tone, in what format
Data = examples, links, input information
CTID is the memory aid: Context · Task · Instructions · Data. The full structure is longer: five sections, Role · Task · Inputs · Expected output · Constraints. CTID's Context is expressed through Role and Inputs; Task stays Task; Instructions become Constraints; Data belongs in Inputs. Expected output is not named by CTID as a separate part.
## Role
You are — [expertise and specialization]
## Task
[Specific goal — one sentence]
## Inputs
[What you're providing: files, URLs, data, context]
## Expected output
[Format, structure, length, quality criteria]
## Constraints
- Don't do: [X]
- Use only: [Y]
- Max: [Z]
<aside> 🧭
Interactive block — on the site: open →
</aside>