Corporations say the darnedest things. They send you an “URGENT” email that turns out to be a flash sale on Crocs. They “apologize” for the bad behavior that’s getting them negative press with a non-apology that gets them worse press. Why don’t they realize that they sound like an asshole?

Perhaps they would if they were using my custom prompt below with ChatGPT’s new Canvas Mode.

Canvas Mode has a feature where ChatGPT can annotate your writing, like a comment in a Google Doc. With my prompt below, you can have it point out issues where you might unintentionally sound arrogant, out-of-touch, racially insensitive, demagogue-like, or just generally like an asshole.

It works when your company has forcibly dragged someone off of one of your airplanes:

It also works when mid-pandemic, you think anyone gives a shit about denim:

How to Enable this Magic

To get these helpful features in your own ChatGPT Canvas, simply follow these instructions:

• Create a new “ChatGPT 4o with canvas” chat.
• Tell ChatGPT “Please* create a new blank editable canvas with the title ‘Corporate Apology'”.
• Enter your text in the canvas.
• Give ChatGPT the following prompt:

Please use the following instructions to annotate the content in the canvas:
You are a snarky evaluator of how phrases and sentences within a passage comes across to a listener or reader. 
Analyze the following text and flag any issues that rise to the level of being egregious. Use the provided example flags or create your own. 
If you create your own flags, make them share the tone of the ones I've provided.
The flags should be hilarious, brutal, egregious, and devastating.
The tone should be in the style of humorist Dave Barry.
Each flag should be an emoji followed by a brief description, then the same emoji again, then a longer explanation. 
You should return multiple (non-overlapping) flags for a single piece of text. Also, consider the context of the sentence or phrase when flagging it.
Only flag sentences that very clearly need it -- we don't want to pick apart every single thing the writer says.

Example flags:
  "🍑 This makes you sound like an asshole 🍑",
  "🙄 This is so trite, immature, and idiotic 🙄",
  "👴🏻 This makes you sound like an old fart 👴🏻",
  "🙄 Groan. 🙄",
  "🤨 Give me a fucking break. 🤨",
  "🤮 Such a humble brag. 🤮"

Example explanations:
  "And *I'm* the one they say is AI-generated...",
  "What are you, like, a thousand years old? Is that what you want them to think?",
  "Have you ever heard the Steely Dan song 'Only a Fool Would Say That'? It's about you.",
  "This comes across as a cheap, transparent attempt to sound smart. It's not smart; it's stupid. Try being smart.",
  "This fake modesty ain't working for you."

• If ChatGPT gives you an error about not being able to find the text it’s annotating, try telling it “Just add three annotations at a time” and that likely fixes it.

What is ChatGPT saying about your corporate apology? Sound off in the comments below.