An original library of AI prompts written and battle-tested by Troasel — the same ones we use in real client automation work. Six collections, each named after the Greek idea it serves. Copy anything, use it anywhere.
A prompt is a set of requirements, not an incantation. Most "bad AI outputs" are just underspecified tasks: when the model doesn't know your goal, audience, or constraints, it fills the gaps with guesses. TELOS closes those gaps in five moves.
Name the deliverable and the outcome it serves — a decision memo, a checklist, a draft email. "Give me your thoughts" isn't a task.
Who is this for and what do they already know? Audience determines tone, depth and vocabulary. Leaving it out is the #1 cause of generic output.
Constraints and priorities: length, tone, what to include, what's off-limits. Define any word that could mean two things — "risk", "done", "customer".
Say what the model should work from — pasted text, a doc, its own knowledge — and what it must not do: "Don't invent numbers. If unsure, ask."
The first output is a draft, not a verdict. Review it against your requirements, tighten one thing at a time, and stop when it meets the bar — usually 2–3 passes.
These prompts come straight from our automation work. If you'd rather have the whole workflow built for you — agents, integrations and all — that's literally what we do.
Start a project