Archives 2026

Stop Rewriting Emails: A ChatGPT Workflow That Sounds Like You

Disclosure: This post may contain affiliate links. If you click and buy, I may earn a commission at no extra cost to you.

Stop Rewriting Emails: A ChatGPT Workflow That Sounds Like You

This is a longer, story-driven guide you can use as a template—not just a list of tips.

A friend texted me: “Can you help me figure this out?” and my brain immediately tried to write a 12-step plan.

Then I paused and asked: what would a simple plan look like?

That tiny reframe changed everything about how I work.

What GPT-5.2 is aiming to improve

The big practical shift: GPT-5.2 is positioned for long-context, multi-step projects—where you plan, execute, and review in one sustained workflow. If you’re building content systems, agents, or repeatable processes, that’s where you feel it most.

The “project thread” rule

Use one thread per project. Don’t mix unrelated tasks. Keep a running summary at the top of the thread so the model stays aligned.

The 3-phase loop: Plan → Execute → Verify

  • Plan: 7 steps, each with a deliverable and definition of done.
  • Execute: do one step at a time; insist on a specific output format.
  • Verify: ask for risks, assumptions, and what to validate.

Copy/paste prompt pack for GPT-5.2

You are my project partner.
Project: [project].
Goal: [definition of done].
Constraints: [time/budget/audience/tools].

1) Make a 7-step plan. For each step: deliverable, time estimate, risks.
2) For Step 1, ask 7 clarifying questions.
3) After I answer, produce: (a) deliverable, (b) review checklist, (c) risk list, (d) next step.

Instant vs Thinking vs Pro (simple decision tree)

  • Instant: you already know what you want; you just need a fast draft.
  • Thinking: you’re unsure and need structured reasoning (tradeoffs, plans).
  • Pro: you’re shipping a multi-part deliverable and want fewer mistakes.

Where people get stuck

  • They ask for a full project in one go (too big, too vague).
  • They skip clarifying questions (hello rework).
  • They don’t force an output format (so results become messy).
  • They never run a review checklist (so mistakes slip through).

Five high-ROI use cases

  • Build a content calendar + internal linking plan in one sitting.
  • Turn research notes into a clean, publishable HTML draft.
  • Draft a spreadsheet spec (columns, formulas, checks) for a side hustle.
  • Create a step-by-step SOP (standard operating procedure) for repeated tasks.
  • Convert a rough project idea into milestones, risks, and next actions.

Recommended tools (affiliate links)

FAQ

  • Should I trust it for facts? Use it for structure; verify facts when they matter.
  • How do I reduce hallucinations? Require assumptions + verification steps.
  • How do I keep consistency? Maintain a project summary and reuse templates.

One-day challenge (do this today)

  • Pick one real task that’s been hanging over your head.
  • Set a 15-minute timer and run the template from this post.
  • Ship a “version 1” (even if it’s imperfect).
  • Write down one thing you’ll improve tomorrow.

Small reps beat big plans. Tomorrow, repeat with a slightly better prompt or checklist.


Related topics: email writing, tone, templates.

If you found this useful, bookmark it and reuse the templates. The real payoff is repetition.

How to Turn a Messy Idea Into a Clear Plan With One “Clarify & Commit” Prompt

Disclosure: This post may contain affiliate links. If you click and buy, I may earn a commission at no extra cost to you.

How to Turn a Messy Idea Into a Clear Plan With One “Clarify & Commit” Prompt

This is a longer, story-driven guide you can use as a template—not just a list of tips.

I once spent an entire afternoon “researching” and ended the day with… more tabs.

That’s when I realized my real problem wasn’t effort—it was decision friction.

I needed a repeatable way to turn fuzzy ideas into a clean next step.

The core idea

ChatGPT is most useful when you stop treating it like a search engine and start treating it like a collaborator. Collaboration needs structure: a goal, constraints, and a definition of done.

The 4 building blocks of a great prompt

  • Goal: what you want the model to produce.
  • Context: what it must know to do it well.
  • Constraints: limits that prevent rambling (length, tone, audience).
  • Format: the exact shape of the output (HTML, bullets, table).

Copy/paste master prompt

ROLE: You are my assistant.
GOAL: Create [deliverable].
CONTEXT: [background, audience, constraints].
FORMAT: [HTML/bullets/table].
QUALITY: Ask 3 clarifying questions first. Then give (1) quick option, (2) best option.
VERIFY: List assumptions and what to double-check.

Worked example (email / blog / plan)

Here’s an example you can adapt. Notice how the constraints do most of the heavy lifting.

ROLE: You are my editor.
GOAL: Rewrite my draft so it sounds like me.
CONTEXT: Friendly, practical, slightly playful. Audience: beginners.
FORMAT: Keep headings + bullets. 900-1200 words.
QUALITY: Reduce fluff, add a checklist, keep it actionable.
VERIFY: Flag anything that sounds like a claim without evidence.

The 15-minute workflow (repeatable)

  • Minute 1: define the goal in one sentence.
  • Minute 2: add 3 constraints (audience, tone, length).
  • Minutes 3-5: let the model ask questions; answer briefly.
  • Minutes 6-10: get the draft in your chosen format (HTML).
  • Minutes 11-13: ask for a checklist + common mistakes.
  • Minutes 14-15: ask for a shorter version you can post on social.

Quality control: the anti-hallucination mini-checklist

  • If the answer includes numbers, names, or dates: ask where they came from.
  • If it sounds too confident: request uncertainty and alternatives.
  • If it recommends actions: ask for risks and safeguards.
  • If it writes in your voice: provide 1-2 examples to match.

Mini prompt library (save these)

  • Turn notes into a draft: “Use my notes. Ask questions first. Then write HTML with headings.”
  • Make it clearer: “Rewrite 20% shorter, keep meaning, keep tone.”
  • Decision help: “Give 2 options and recommend one with tradeoffs.”
  • Make a plan: “7 steps, each with deliverable + risks + next action.”
  • Make it publish-ready: “Add meta description, title variants, and internal link ideas.”

Recommended tools (affiliate links)

FAQ

  • Do I need long prompts? No—just the right context and constraints.
  • How do I keep outputs consistent? Save templates and reuse them.
  • What should I verify? Any factual claim, statistic, or specific recommendation.

One-day challenge (do this today)

  • Pick one real task that’s been hanging over your head.
  • Set a 15-minute timer and run the template from this post.
  • Ship a “version 1” (even if it’s imperfect).
  • Write down one thing you’ll improve tomorrow.

Small reps beat big plans. Tomorrow, repeat with a slightly better prompt or checklist.


Related topics: prompting, planning, focus.

If you found this useful, bookmark it and reuse the templates. The real payoff is repetition.