How to Write a Code of Conduct With AI (And Still Sound Like a Human)🧍♀️🤖
This is the first in a series of practical articles on how to use AI in real-world HR work. While this one walks through writing a code of conduct, the process and principles apply to any kind of policy—remote work, expense guidelines, onboarding docs, you name it. If you’re building policies across borders, roles, or contract types, this method will work for you.
TL;DR:
You need a code of conduct. Fast. One that applies across employment types, makes sense in every country you're in, and doesn’t read like a legal threat. Here’s how to draft a policy that people actually understand and follow—with help from AI and some honest human filters.
⚠️ Reminder: Never include personal, private, or confidential information in AI prompts. Assume everything you type could be shared. Keep it safe and professional.
We start with a persona—like "HR at Company X"—so we can ground the context in real company conditions. This makes the AI prompt much more useful (and less likely to generate something generic).
You: HR at Company X (100-person startup)
Your mission:
Write a code of conduct that applies to employees, contractors, freelancers, and EOR hires across Europe—without sounding like you're policing everyone’s vibes.
You have:
100 people, fully remote 🌐
Multiple contract types (direct hire, freelance, EOR) in different countries across Europe 📄
A remote culture that values autonomy 🤝
A need for consistency, clarity, and zero ambiguity ✉️
Let’s go.
1. Know What You’re Solving 🤔
Before you ask AI to help, define the real problem.
Business problem:
“We want everyone—no matter their contract or country—to understand the behaviors we expect, what happens if things go wrong, and how we protect psychological safety without becoming a compliance robot.”
What this policy needs to cover:
Who it applies to (everyone, regardless of employment type)
What "good behavior" looks like at Company X
What’s out of bounds (harassment, discrimination, etc.)
How to report concerns
What happens after a report
How we ensure fairness across contract types and geographies
Write this down before you brief the bot.
2. Write a Smart Prompt 🧠
Don’t just say, "Write me a code of conduct." Be specific. Like this:
ROLE: You are an experienced HR policy writer. You have vast experience with distributed, remote-first startups.
CONTEXT: Company X is a 100-person remote company with staff across Europe. People are employed under direct contracts, freelance agreements, or via an Employer of Record. We need a single code of conduct that applies to all.
COMMAND: Draft a clear, inclusive code of conduct for this context.
REQUIREMENTS:
1. Define who the policy applies to
2. State expected behaviors and values (you can upload or copy/paste your values)
3. List unacceptable behaviors
4. Outline how to raise a concern
5. Describe how complaints are handled and who handles them
6. Use language that works across cultures and legal systems
QUESTIONS: ()
- What’s missing that might confuse people?
- Where could cultural context create misunderstanding?
- How do we make this feel fair to all contract types?
- What concerns would people have about retaliation or bias?
STYLE: Clear. Inclusive. No legalese.Why this works:
Role = helps the AI think more like a real HR partner 🧠
Context = avoids random corporate speak and prevents hallucination and generic answers
Command = keeps it focused
Requirements = covers your bases; force the AI to cover all bases and not “forget” elements.
Questions = stress-tests it for reality
The "QUESTIONS" section in the AI prompt structure serves a very strategic purpose:
It tells the AI to think critically before it writes. Instead of blindly generating text, these questions act like a mini pre-mortem—forcing the AI to:
Anticipate confusion or resistance from your audience
Spot cultural, legal, or contractual pitfalls
Think about fairness and emotional impact (especially across diverse contract types)
Raise red flags before they become real-world HR headaches
In short: it helps the AI simulate the kind of second-guessing and stress-testing a seasoned HR pro would naturally do. It makes the output less generic, more thoughtful, and more aligned with your org’s reality.
Style = keeps it human. Saves you from getting output that sounds like it was written by a robot in a pinstripe suit.
3. Edit Like a Cynical Engineer 🧥
AI’s first draft will be 70% decent, 30% cringe.
Here’s how to fix it:
Check for any policy language that sounds like it was written for a California megacorp
Translate vague values like "integrity" into actual behaviors people can spot
Add examples: "If someone jokes in a team call about another person's accent, that’s not okay—even if they didn’t mean harm."
Cut any section that feels like it belongs in a court deposition
Bonus tip: both ChatGPT and Google Gemini offer a free "Canvas" feature (as of this writing). Ask the AI to place your draft in Canvas—so you can edit sections easily, leave comments, collaborate with others, and export clean versions without toggling between tools.
4. Run the Legal + Logic Filter 📖🤦♀️
Before rollout:
Check basic compliance (but don’t aim for full legal review in every country—just flag obvious risks)
Make sure it’s clear this applies to everyone
Test tone: Does it make people feel supported, not punished? 🤝
5. Simulate the Reactions (AI-Powered Pre-Mortem) 🥳
Ask AI to act out team reactions. Prompt it like this:
ROLE: You are team members at Company X.
CONTEXT: You've just been sent a draft of the new code of conduct.
COMMAND: React to the policy as real people with different contracts and cultural backgrounds.
REQUIREMENTS:
- Play these roles:
- Freelancer in Spain who worries about being left out
- EOR hire in Romania curious how to report an issue
- Longtime employee in Berlin who values transparency
- Share each person’s:
- First reaction
- Questions or worries
- Slack message they'd post if confused
STYLE: Informal, realistic, honest.Ask questions like:
“What might frustrate you about this policy?”
“Where do you see unfairness?”
“What would you ask HR in Slack after reading this?”
Document the reactions. If the AI (as your team) raises the same issues you dread hearing in real life, fix them now—not during your all-hands Q&A.
6. Adjust. Humanize. Ship It. 🚀
Use what you learned to:
Rewrite anything unclear
Add a friendly FAQ
Make sure "how to report something" is actually easy
7. Communicate 💬
Don’t drop a PDF in Confluence and call it a day.
Have AI help you create:
Slack post (quick summary + link + Q&A thread)
Email (explains what’s changing, why it matters)
Confluence (Internal site) page (policy + FAQ + contacts)
Tell people what to expect. Make it feel human.
Bonus Round: Build Your Own Policy Bot 🌟
If you're drafting policies regularly, make a custom AI trained on:
Your structure
Your tone
Your org's quirks
Now any HR person can say: "Create a contractor onboarding policy" and get a usable draft that doesn’t need an emergency rewrite.
Curious? Stay tuned—I'll be sharing a step-by-step guide on how to build your own custom GPT or Gemini Gem to automate policy drafting soon. It's easier than it sounds.
Want more like this? Hit subscribe or forward to your favorite HR person who’s knee-deep in international contractor weirdness.