AI & No-Code Guidelines

We’re in the exploration phase (April 2025)

We’re still discovering how AI and no-code tools can best support our design work.

It’s absolutely fine if you’re not using them yet—focusing on discovery, user research, or hands-on design is just as valuable.

For those experimenting: be curious. Be open-minded. Try things out, share what you learn, and don’t be afraid to make mistakes. This is a space for learning, not perfection. Yet :)

🚦Quick Principles

• Think privacy-first.
• Use AI to accelerate (not replace) your thinking.
• Choose the right tool for the right task.
• Keep things simple, safe, and explainable.

---

✅ Do’s

🔐 Keep it secure

• Use only approved tools (on MCP or whitelisted).
• We are currently looking to license Loveable, but this might change
• MCP is being developed by our Innovations team
• Strip or anonymize all customer data, logos, images, and content.
• Never upload API tokens, credentials, internal links, or Figma access tokens.
• When in doubt, ask in #design or #infosec channels.
• Have you found a new tool that may be beneficial? Consult it on one of the channels above.

🛠 Use no-code when it makes sense

• Great for dynamic forms, inputs, logic, and dropdown flows.
• Rule of thumb: If a prototype takes 1 hour in Loveable vs 1 day in Figma → go for it.
• Use tools like V0 or Loveable to test flows, not to ship production UI.

🧠 Think before you click

• Spend 10 minutes scoping before you jump into any tool.
• Choose a tool based on what you want to learn, not what’s trendy.

👯‍♂️ Combine strengths

• Use AI for naming, copywriting, first drafts, or quick explorations.
• Use Figma and your brain to refine and validate.

🤝 Challenge developers (constructively!)

• Use AI or no-code tools to quickly prototype an interaction that’s hard to explain in words.
• Suggest alternative flows and ask: “Which one feels more feasible?” → helps you align faster.
• Show what might be possible without custom code—as a proof of concept or fast MVP.
• Ask: “If we had to ship this tomorrow, what’s the simplest version?” → AI can be a great conversation starter.

📎 Document what AI helped with

• Leave a short note in your spec or prototype: "First draft created with ChatGPT," etc.
• Use reverse prompt engineering to instruct the tool to create a description of the prototype.

---

❌ Don’ts

🚫 Don’t upload sensitive stuff

• No client logos, real screenshots, user data, or internal documents
• Never upload API tokens, credentials, internal links, or Figma access tokens.
• Be very cautious of tools with unclear license terms (e.g., V0 has a “grant-back” clause: they can reuse what you upload).

🧼 Don’t over-polish AI outputs

• Prototypes from AI tools are disposable.
• Don't waste time trying to make them look "on brand." Use them for speed & testing.

📦 Don’t trust defaults

• Always double-check generated outputs for accessibility, logic, and inclusivity.
• AI often creates visually plausible but functionally broken components.

👁 Don’t use AI as your only lens

• Always validate ideas with your team, users, or devs.
• AI can’t replace critique, feedback, or context.

---

👀 No-Code tools watch-outs

---

🔮 Future?

• Expect better Figma integrations soon – AI won’t stay in separate tools.
• Treat AI tools as experimental partners, not production tools (yet).
• Document learnings. Share wins and failures in the team. We're all figuring this out.