🏴‍☠️ 9 AI rules before you paste anything

Almost every company says it "uses AI" now. Ask how the team is supposed to use the thing, though, and you'll get a shrug, because nobody ever wrote the rules down and people just paste sensitive work into chat windows and hope it lands. That gap is the whole problem, and it is bigger than most leaders admit.

This memo caught my eye because someone finally closed that gap. The author leads a team, got tired of watching smart people guess, and wrote a short instruction they made non-negotiable: nine rules, one page, no exceptions. To me it reads like the AI onboarding doc most offices never bothered to write.

Here is why I think it is worth your time: it protects the two things that burn companies fastest, your private data and your credibility, without sliding into a lecture. I have laid out all nine steps below, each paired with the reason it exists. You can lift the whole framework for your own team.

_*Ad

The first rule kicks in before you open a chat. Ask yourself one question. Would you be fine seeing this text in the company-wide channel, with your name on it?

If not, swap names for roles, so "Sarah at Acme" turns into "the client." That small edit takes five seconds and strips out most of the exposure.

Two clicks matter more than people realize. Never hit thumbs up or down on a work chat, and never hit share. The creator warns that the feedback button can ship your conversation off to train the model, so one careless tap can expose client details you never meant to share.

Connectors are the same danger at a larger scale. A single integration can read your whole inbox, your full Drive, your entire Slack, and you will touch almost none of it. Every connection you add is one more door your data can walk out of.

Connect only what earns it. Prefer read-only. Review it every month.

Here is the rule that belongs on every laptop: if you send it, you wrote it. The original poster is blunt about verification, pushing you to check every number, every name, and every source before it leaves your hands.

Why that matters: AI will invent a statistic that sounds real, complete with a plausible decimal and a confident tone. Drop it into a client deck, and the model takes zero blame. You will, in front of the room.

So label raw output when it is raw, with a tag like "Claude suggested this, worth exploring?" Unedited AI text turns your work into someone else's, and your manager is not your proofreader. That small tag keeps trust intact.

The Dell Pro powered by Intel® Core™ Ultra with Intel vPro® adjusts power usage based on how you work, so your battery life never slows you down.

Explore More

This is the rule that surprised me most, and I think it is the smartest one here. Before you open any tool, write down three options of your own. Only then bring in the model, with a prompt like "What did I miss, and how do these compare for our situation?"

One detail the expert insists on, never end a prompt with "right?" or "maybe?" Those endings beg the model to agree, and an AI that nods along will never catch your mistakes.

Start from your own thinking, and it sharpens you instead of replacing you.

The same logic drives the daily habit the memo calls the brain gym: write one draft by hand, run one calculation yourself, make one decision before you ask. We all did this before ChatGPT arrived, and skills you stop using fade faster than you expect. The loss stays invisible until the day you reach for one and find it gone.

This is my favorite framing in the whole memo. Claude does not know your clients, your jargon, or what happened in Tuesday's meeting, so tell it. Context is the cheapest upgrade you have.

Send a screenshot instead of retyping numbers, and ask it to open every relevant email, doc, and tab before it starts.

Most weak output comes from missing context, not a weak model. When quality drops mid-conversation, the fix is simple: open a fresh chat. Long threads collect clutter that drags the answer down.

Feeding Claude real context is a reflex, and reflexes come from reps, not from reading a memo once. The claude track in 3 Minute AI runs that habit as short daily lessons, with a built-in lab where you write the prompt instead of nodding at a summary. Daily lessons and two full courses are free, so it costs nothing to see if it sticks.

Early-stage teams move quickly, but their internal systems often struggle to keep up. In the first few months of building a company, it is easy for documentation, roadmaps, meeting notes, hiring plans, and customer insights to scatter across different tools and conversations. At first, that mess feels manageable. But as the team grows, every unclear process becomes a drag on execution. The most effective founders treat their workspace like a product. That means designing it intentionally, with a clear structure, useful templates, shared naming conventions, and a simple onboarding experience for every new team member. A strong workspace should make work easier to find, decisions easier to track, and priorities easier to understand. Operational debt compounds just like technical debt. Every duplicate doc, lost decision, or unclear handoff slows the team down. By building a clean, scalable workspace early, founders create a stronger foundation for faster execution, better collaboration, and less time wasted searching for context. Build a workspace that scales with your team.

Try Notion today and get up to 3 months free of Notion Business with Notion AI.

Then keep it short. If a two-line prompt produced a two-page doc, send the two lines.

People spot those AI tells fast. Strip "it's not X, it's Y," "delve," "seamless," and the pep-talk endings, because trust dies the moment readers notice them.

The last rule ties the rest together. Every prompt should carry three things: the goal, the context, and a clear picture of what "done" looks like. One sharp ask beats five rounds of "no, not like that," and you can delete "think step by step," since modern models reason without the nudge.

Before your next work prompt today, swap every client name for a plain role and jot down three of your own options first, so the model fills gaps instead of inventing facts.

If you want all nine rules laid out with the reasoning behind each one, read the full nine-rule memo.

Worth 10 minutes if your team is pasting work into AI faster than anyone has written down the rules for it.

Hit reply and tell us why.