Table of Contents
- UX designers are using AI wrong
- What is a personal AI operating system for UX designers?
- Why most UX designers are using AI wrong
- How to codify your UX expertise before building AI agents
- How to map your UX judgment into an AI agent
- What kind of AI agents are useful for UX designers?
- What AI is changing about the UX designer's role
- How to get started with your personal AI system this week
- The bottom line
Do not index
Read time: under 8 minutes
UX designers are using AI wrong
Ask most UX designers what they use AI for and you'll hear the same things.
Drafting copy. Summarising research. Rewriting a brief. Small tasks. One-off jobs. Nothing that changes how much time they spend on the hard stuff.
That's because they're using AI like a fancier Google search.
But when it's built properly, AI stops being a task tool and starts being a super powerful thinking partner.
In a livestream, Chris — Founder of UX Playbook — sat down with Ryan Rumsey, founder of CDO School, author of Business Thinking for Designers, and one-on-one coach to senior design leaders, to talk about what building a real personal AI operating system looks like.
Ryan has spent years codifying how senior designers think — long before AI existed. What he shared was practical, a little uncomfortable, and exactly what most designers need to hear.
Let's get into it.
👉 The insight in this blog traces back to a thoughtful conversation with Ryan — here’s the full discussion:
What is a personal AI operating system for UX designers?
A personal AI operating system is not software. It's not a product. It's the codified version of how you think — your decision-making logic, your judgment calls, your way of navigating the messy situations that come up in design work, made legible enough that an AI agent can actually use it.
Ryan puts it plainly:
"They're experts in that. But then they are amateurs at explaining that to other people — even themselves."
Why most UX designers are using AI wrong
Here's what actually happens when most designers sit down with an AI tool.
They type a vague question. They get a generic answer. They feel vaguely disappointed. They tell themselves AI is overhyped. They go back to doing the work themselves.
Ryan sees this constantly. And his diagnosis is sharp:
"These models are going to predict no matter what. And if you don't understand yourself explicitly what is good or what is bad, you're just going to be in this circle of frustration."
He compares it to walking up to a senior executive and asking: "Give me your strategy." And getting... nothing useful. Not because the exec doesn't know things. Because they've never had to make their thinking explicit.
That's the same gap most designers have with their personal AI system. The problem isn't the model. It's the input.
How to codify your UX expertise before building AI agents
Here's the thing most AI guides skip entirely.
Before you build anything — before you write a single prompt, before you open a new chat window — you need to do this one thing: figure out how you actually think.
Ryan wasn't doing this for AI originally. He started years before large language models existed. He was just trying to get better at mentoring.
He'd hired a brilliant designer onto his team. Brought them into executive meetings. And one day that person said something that stopped him cold:
"I understood everything in that meeting. I knew what you all were talking about. But I don't know how I go from where I am in my career today to where you are."
That moment forced Ryan to look at the gap between giving advice and giving guidance. Advice sounds like: "You need to be more strategic." Great. Now what? Guidance actually shows someone the path.
His approach to closing that gap: stop thinking in skills, start thinking in situations:
"What are these situations that are common? Have I faced them before? If yes, how did I fix them before? Great — here's now a technique or approach I used."
Practical exercise for UX designers:
Write down five recurring situations from your work: he budget just got cut, a stakeholder pushed back on your direction, a junior asked you to review their work, etc.
For each one, answer: What triggered it? What did I do? Did it work? What would I do differently?
That exercise is the foundation of your personal AI system. It's not a skill framework. It's a situational map of your judgment.
Join 13,053+ Designers for FREE weekly UX Insights
Every Wednesday, I send out 1 actionable framework to grow your UX career 🌱 — No fluff. Always 2 minutes or less.
How to map your UX judgment into an AI agent
Once you've done the work of codifying your thinking, feeding it to a model is surprisingly straightforward.
Ryan's method involved a mix of resources he already had — SRT files from recordings, logic trees built in FigJam, quick voice memos. He'd give all of it to Claude with explicit instructions.
"I want it to be another version of me that somebody else can use because I'm not always available."
You're not building a chatbot. You're building a thinking partner that operates on your logic — one that can help someone at 11pm without needing a 45-minute call with you.
The part most people miss: separating building from testing.
"I separate the acts of building the agent from testing the agent."
Here's how it works in practice. Build the agent in one chat. Once it feels stable, open a fresh chat — no memory of the build process — and run it like a real user. Type random scenarios. See what breaks. Log everything. Take that log back to build mode and improve.
It's the same logic as dev branches and test environments. Build mode and test mode are different contexts. Treat them that way.
"It's just a series of Q&A and testing with myself — where I'm sort of acting as user."
What kind of AI agents are useful for UX designers?
Not the obvious ones.
"Make me a document" — fine. But that's not where the real time savings are.
The agents worth building are the ones that handle things that currently require you, specifically, in a room, for 30 to 45 minutes.
"The things that I do repeatedly that require my judgment, require my learned knowledge — and don't have a clear sort of tree structure — those are the things where I go: that usually takes a lot of my energy."
Ryan's example from CDO School: one of the most common questions he gets is "how do I talk to executives?" So he built a skill around it. The agent asks the person to describe their situation, identify their stakeholders, name any hard constraints. Then it routes them to the exact lesson or resource that addresses it.
No 45-minute call needed. The agent runs on his logic, 24 hours a day.
For UX designers, think about this 3 questions:
- What questions do you answer on repeat?
- What situations keep landing on your desk?
- What things can only you explain — but you're explaining them over and over again?
That's your agent backlog.
What AI is changing about the UX designer's role
Before you panic: AI is not coming for your whole job. It's coming for the execution part. Which means the value shifts from doing to directing.
Ryan uses a TV analogy that's genuinely useful. Think about the showrunner role. Not the director of a single episode. Not the writer of one scene. The showrunner is the person who holds the story together across the entire series — making sure every beat connects, every scene serves the bigger arc.
That's what AI needs from UX designers now. Someone who can step back and say: this doesn't fit the story.
And there's a sharp observation from Ryan about why creative people often feel worse about AI than developers do:
"For creative people, AI seems to be the work that fills them with joy and love and humanity — and leaves them with drudgery. Whereas for coders, AI takes the drudgery of writing code and allows them to be more of that creative director, that architect."
The honest question every UX designer needs to sit with: what part of your work do you actually love? Because that's probably what AI should not be doing for you.
The meetings. The documentation. The repetitive stakeholder write-ups. That's your drudgery. Hand it over.
The thinking. The judgment. The creative direction. Keep that.
How to get started with your personal AI system this week
You don't need to be a developer. You don't need terminal. You don't need a six-month project.
You need one thing: a repeating frustration.
"If you have ever found yourself asking the same question into a chat window multiple times — that's your way in."
Ryan recommends Claude's co-work feature as the practical starting point for designers who aren't engineers. Instead of feeding documents into a chat, co-work gives Claude access to a folder on your machine. It creates documents for you. Builds structure. Populates your system.
The mental shift is important here. You're not the user. You're the expert. Claude is the consultant you've hired to make things.
"You can kind of be more of the expert in the room — and you've hired Claude to make you things."
Three steps to start this week:
Step 1. Identify one situation you navigate on repeat. Write down how you handle it. Even rough notes work.
Step 2. Give that to Claude — transcript, voice memo, bullet points, anything. Ask it to help you turn it into a decision framework.
Step 3. Test it in a fresh chat. See where it breaks. Fix one thing. Repeat.
That's it. That's your first piece of personal AI infrastructure as a UX designer.
The bottom line
Building a personal AI system is really just the process of making that judgment legible. Not for AI's benefit. Yours. When you can explain how you navigate a situation — step by step, with criteria for what good looks like — you become a better mentor. A better leader. A better designer.
AI just gives you leverage on top of that.
Do the unglamorous work first. Sit with your own thinking. Map your recurring situations. Write down what you do — not what you think sounds impressive, but what you actually do. That's your raw material.
Then build the system around it.
Ryan built his before AI even existed. That's the point. The tool is secondary. The clarity of your own thinking is what makes it work.
Start with one situation. One recurring moment where you always know what to do — but couldn't quite explain why. Write it down this week.
That's the start of your personal OS.
Whenever you're ready, there are 4 ways I can help you:
1. Junior Designer Bundle: Transition to UX with the ultimate handbook (120+ videos, 80+ templates, 75+ examples) to craft an unforgettable portfolio & get hired.
2. Senior Designer Bundle: Become a design leader with systems to help you build a meaningful career & grow your designers. Join 500+ aspiring leaders.
3. UX Portfolio Critique: Get a 30-minute video review of your portfolio. A checklist of actionable things to fix, in less than 48 hours. Get a personalised portfolio critique here.
4. Job Sprint Course: Get battle-proven frameworks and interactive workshops to: build a memorable personal brand, a killer strategy for job applications, and tactics to nail job interviews. Get hired in UX with Job Sprint.
