🤖 I quit doing these 3 things (AI does them now)
There’s a difference between using AI to help with a task and quitting the task. Quitting means it leaves your plate entirely. You stop being the writer and become the editor, and you keep only the checkpoint. Here’s what I actually quit, with the honest fine print on each.
1. I quit writing first drafts. Newsletter sections, video scripts, outlines. Every single one starts as an AI draft now, from a template that includes my voice rules and a sample of my writing. The prompt is in today’s kit word for word.
- The guardrail: I do the final voice pass myself, and I read everything aloud before it ships with my name on it. The draft is delegated. The judgment never is.
- The honest miss rate: some drafts get heavily rewritten. A few get thrown out completely. It’s still hours faster every week, because editing is faster than staring at a blank page. (For the task-by-task version of those hours, see the AI Work Efficiency Kit.)
2. I quit doing research legwork. Gathering, comparing, checking. Last week’s issue is the live example: I had AI verify every pricing table against the official pages, because half the articles out there are quoting 2024 numbers with a 2026 date on top.
- The guardrail: it’s built into the prompt. A source link for every claim, “I couldn’t verify this” is an acceptable answer, and I spot-check the numbers that matter before they reach you.
3. I quit reformatting my own content. One finished piece becomes the X post, the LinkedIn post, and the caption drafts. AI does the reshaping. I do the choosing.
- The guardrail: nothing posts unread. Every version gets my eyes before it gets an audience.
What I never delegate: final decisions, anything I’d have to apologize for if it went out unreviewed, and replies to you. When you email [email protected], that’s me, every time.
The pattern across all three: delegate the doing, keep the judging. If checking the result is faster than doing the task, the task is a candidate. That’s the whole test, and the kit below walks you through it for your own week.
🎁 🎯 Free download: The AI Delegation Kit
The quit-or-keep decision tree (4 questions that tell you if a task is ready to hand over), all 3 delegation prompt templates with the guardrails built in, the honest miss-rate log, and the Friday 10-minute delegation review. Works with ChatGPT, Claude, or Gemini.
Frequently asked questions
What tasks should you delegate to AI?
Use one test: if checking the result is faster than doing the task, the task is a candidate. The three big ones from my own week: writing first drafts, research legwork (gathering, comparing, verifying against official sources), and reformatting one finished piece into social posts and captions.
What should you never delegate to AI?
Final decisions, anything you would have to apologize for if it went out unreviewed, and personal replies to real people. The pattern: delegate the doing, keep the judging — you stop being the writer and become the editor, but the checkpoint stays yours.
Does AI-delegated work still need human review?
Yes — every time. The guardrails that make delegation safe: a final voice pass on drafts, reading everything aloud before it ships, requiring a source link for every research claim (with "I could not verify this" as an acceptable answer), and never posting a reformatted version unread.
Is delegating first drafts to AI actually faster?
Honestly, with a miss rate: some drafts get heavily rewritten and a few get thrown out completely. It is still hours faster every week, because editing a draft is faster than staring at a blank page.
