Someone sends you a meeting prep doc the night before a call.
You open it expecting a quick read. Instead, it's six sections, 2,500 words, a stakeholder overview, background, open questions, recommended talking points, and a summary of every conversation that's ever happened. Your meeting starts in 30 minutes, and now you have homework.
This has become one of the quieter frustrations of working in an AI-enabled workplace. It's not that people are using AI. I hope they are. It's that too often, the first draft becomes the final draft.
The sender saved themselves 20 minutes, but now five other people each have to spend 10 minutes figuring out what matters.
AI has a tendency to sound a little more polished than most people naturally write. Unless you tell it otherwise, it'll often reach for bigger words, longer sentences, and a more formal tone than necessary. Sometimes that's the AI. Sometimes it's us asking it to "make this sound professional."
Either way, the result is usually the same. You end up reading something that technically sounds good, but doesn't sound like the person who sent it.
This isn't a knock on AI writing. I used AI to help me write this blog because it helps me communicate my ideas more clearly. My natural writing is functional, not elegant, and I'm perfectly happy letting AI improve the wording. The thinking is still mine. That's where the value comes from.
We've all done it. You throw AI a quick prompt, skim the response, think, "Looks good," and hit send. Sometimes that's fine. A lot of times it isn't.
AI is great at giving you everything. The problem is that the reader rarely needs everything. They need the few things that matter, and AI can't decide that for you.
Meeting prep docs are my favorite example. A good prep doc doesn't try to tell me everything you know about the account. It tells me what I need to know before the meeting. What's the objective? Who should I pay attention to? What's changed since the last conversation? What should I avoid? Everything else can be reference material if I need it.
That's where people get AI wrong. The value isn't in generating a document. The value is deciding what deserves to be in it. AI doesn't know what's important, what your audience already knows, or what can safely be left out. That's still your job.
Skip that step and you've just turned a five-minute read into a scavenger hunt for what’s important.
The easiest way to avoid the problems above is to stop thinking about AI output as a finished product. Treat it like a first draft. Before you hit send, put yourself in the reader's shoes. Did you make the important information easy to find? Did you remove the things that don't matter? Did you add your own judgment, context, and perspective?
The goal isn't to make AI write something that looks polished. The goal is to create something useful for the person receiving it.
Before I send something important, I use this prompt to pressure test it from the reader's perspective. It doesn't replace a human review, but it helps me catch the things that are easy to miss when I'm too close to the work.
Act as a thoughtful editor reviewing this document from the perspective of the person receiving it. Do not rewrite it yet. First, help me identify where it can be improved.
Review it across these areas:
1. Clarity
2. Relevance
3. Reader experience
4. Voice and authenticity
5. AI-generated patterns
After reviewing, provide:
Do not rewrite the entire document unless I ask you to. Focus first on improving the thinking, prioritization, and usefulness of the content. Your role is to help me make better decisions, not simply make the writing sound better.
Not sure on your next step? We'd love to hear about your business challenges. No pitch. No strings attached.
