Skip to main content

Documentation Index

Fetch the complete documentation index at: https://help.pantaos.com/llms.txt

Use this file to discover all available pages before exploring further.

Prompting is how you communicate with AI. Good prompts get good answers; vague prompts get vague answers. A few principles take you 80% of the way.

​
The core pattern

A solid prompt has four parts:

Role

“You are a senior B2B copywriter.”

Task

“Write a 200-word LinkedIn post announcing our new analytics product.”

Context

“The audience is heads of operations at mid-market firms. The product saves 10 hours per week of manual reporting.”

Constraints

“Use a confident-but-not-bro tone. Avoid jargon. Include one specific outcome metric.”

​
Patterns that work

Including one or two examples of the output you want is the single most effective improvement you can make.
“Return as a bulleted list,” “use a table,” “limit to 100 words” — formats reduce ambiguity.
Paste the source document or attach a knowledge base. Generic prompts get generic answers.
“Now make it shorter.” “Now sound more confident.” “Now use last year’s data.” Iteration beats one-shot perfection.
“Before answering, list the trade-offs you’re weighing.” Forces structured reasoning.

​
Anti-patterns

Vague openers

“Help me with marketing.” → “Draft a one-paragraph LinkedIn post about ___.”

Conflicting constraints

“Be detailed but brief.” Pick one — the model can’t satisfy both at once.

No examples

“Write in our voice” without an example of your voice. Show one.

Polite hedging

“Could you possibly maybe consider…” wastes tokens. Be direct.

​
Prompt templates worth saving

Summarizer

“Summarize this document for [audience] in [format]. Highlight [specific aspects].”

Drafter

“Write a [length] [type] for [audience] about [topic]. Tone: [tone]. Constraints: [constraints].”

Reviewer

“Review this draft against [criteria]. Flag specific issues with line numbers.”

Extractor

“From this text, extract every [thing]. Return as a JSON array with fields [list].”

​
Where prompts live in PANTA OS

The system prompt of an assistant is the most-reused prompt in your workspace. Invest in getting it right.
Each step has its own prompt template. Tune individual steps without breaking the whole flow.
Day-to-day prompting. Learn the patterns above; they pay back forever.
When something doesn’t work, the answer is almost always “give the model more context, more examples, or clearer constraints.” Rarely “switch to a smarter model.”