AI Workflows Library · Practical AI use cases & templates

AI for writing

This page is a hub: practical patterns + copy-paste prompts.

What you will get

The 5-step workflow (stable)

  1. Define purpose and audience.
  2. Provide inputs (facts, links, constraints).
  3. Ask for a structure first (outline).
  4. Generate the draft.
  5. Run a quality pass (see: Quality checklist).

Core patterns

1) Outline first (then write)

Prompt

Create an outline for: <topic>.
Audience: <who>.
Goal: <what outcome>.
Constraints: <length, tone, must-include points>.
Output: a numbered outline (H2/H3).

2) Rewrite with constraints

Prompt

Rewrite the text below.
Keep meaning. Improve clarity.
Constraints: <tone, length, level>.
Keep terms unchanged: <term list>.
Text:
<PASTE>

3) Edit for correctness (fact guard)

Prompt

Review the text for possible factual errors or ambiguous claims.
Mark:
- [CHECK] needs verification
- [OK] likely correct
Then propose safer rewrites that avoid unverified claims.
Text:
<PASTE>

4) Style transfer (without changing meaning)

Prompt

Rewrite in the style: <formal / friendly / direct / technical>.
Do not change meaning or add new facts.
Text:
<PASTE>

5) Summarize for decision

Prompt

Summarize this content for a decision maker.
Output:
1) 5 bullet summary
2) Risks / unknowns
3) Recommended next step (1)
Text:
<PASTE>

6) Turn notes into a structured doc

Prompt

Turn these notes into a structured document.
Format:
- Title
- Context
- Key points
- Open questions
- Action list (Owner / Due / Status)
Notes:
<PASTE>

7) Generate alternatives (A/B/C)

Prompt

Create 3 variants of the same message:
A) short and direct
B) neutral and formal
C) friendly and persuasive
Constraints: <max length, must include X>.
Draft idea:
<PASTE>

Starter packs (copy-paste)

Minimal request

Task: <one sentence>.
Audience: <who>.
Constraints: <tone/length/must-include>.
Output format: <bullets / table / sections>.

Pro request (better control)

Task: <task>.
Context: <background>.
Inputs: <facts/data you provide>.
Constraints: <privacy, compliance, tone, length>.
Output:
1) Draft
2) Assumptions (if any)
3) Quality checks performed

Related use cases