It’s never been easier to learn by doing
Building forces you to learn things you’d never otherwise have to.
When I was at school, I took part in the a programme run by the entrepreneurship charity Young Enterprise. They get children to set up and run their own business, with adult mentors to guide them. You sell shares to friends and family, take on a role and try to make a profit running your company.
Young Enterprise is all about learning by doing, because that’s the best way to learn. You can shadow people, read books, listen to podcasts and watch tutorials, but actually doing the thing is how you truly understand it.
AI allows us to do more and therefore to learn more. Tools like Claude Code and Codex let UX people go beyond their traditional role and actually make the thing they normally just contribute to. With AI changing how we design and build software, the best way to learn about any of it is to just try it.
AI has removed the barrier between designing and building
In our day-to-day work as designers and researchers, we typically operate at a layer of abstraction from the making of the software. We are in service of others who are writing the code or making decisions about it. We don’t ‘get our hands dirty’ building the thing ourselves.
Imagine if we were constructing a building. Engineers are the ones laying bricks, quarrying stone, shaping it into place. Designers are drawing pictures of what the building should look like and handing those pictures to the engineers, but they’re not actually building it themselves. Researchers are talking to the people who will eventually use the building, then advising the people drawing the pictures and the people laying the bricks. Everyone is contributing, but only the engineers are actually constructing the thing.
That’s how software development has always worked. But now tools like Claude Code let you go end-to-end: from idea to working product, without writing the code yourself.
It’s also getting easier to try this stuff out. Recent releases like the OpenAI Codex Mac app and Claude Code inside the Claude Mac app are essentially wrappers for the terminal that make the whole experience feel a bit less intimidating. You don’t have to stare at a blank command line. You can dip your toes in through a more approachable interface and just get started.
Building forces you to learn things you’d never otherwise have to
Over the last couple of months, I’ve been building a web app in my spare time using Claude Code. It takes the emails that schools send parents and extracts the tasks and events so it’s easier to keep track of what you need to do.
I feel like I’ve been learning faster than any time in my career doing this, because building a real thing forces you to take on roles that you don’t normally. When you’re part of a multidisciplinary team, you contribute but when you build something yourself, you are the whole team.
Because it’s just me (and Claude), I’ve had to learn all of the following:
- Modern web stack: understanding how web apps are built and the platforms they use.
- Prompt engineering: writing and evaluating the core prompts that make the app work.
- Technical architecture: designing an email processing system that is resilient and scalable.
- Security: understanding how to protect the app from abuse and keep users’ data safe.
- Brand and visual design: defining how the app feels, choosing typefaces, colours, designing a logo and so on.
- Marketing: understanding what types of content work on Instagram, and filming and editing reels.
- Legals: writing the terms of service and privacy policies.
- Pricing: working out the best price point to sell at and model the commercials over time to ensure it’s sustainable.
Before AI, it would have been too intimidating (and time consuming) to attempt all of that. But now you can have an LLM holding your hand every step of the way. Whenever you don’t know how to do something or you get stuck, you can ask for help. For £20/month, you get a Claude Pro subscription that gives you access to Claude Code and an always-on teacher and safety net.
Most people are still watching from the sidelines
You’d think from reading LinkedIn that everyone is doing this, but from what I can see, only about 10% of people in most organisations are actively exploring and experimenting with AI beyond basic usage of ChatGPT. And only 1-2% are trying out coding tools like Claude Code.
So if you are using these tools to build, it’s putting you in a really good place career-wise. If you’re looking for a job or will be in the future, having experience building products with these tools gives you a significant advantage over other candidates. You’re not just curious about AI, you’re taking steps to understand how it works in practice.
Developers have always had side projects – it’s almost expected of you. But UX people haven’t traditionally had side projects because they could do the research and design the experience, but they couldn’t build the end product themselves. Now you can.
Just get started
My advice to pretty much anyone in UX is to get a side project. Think of something annoying in your day-to-day life and build a little app around it. It doesn’t need to be polished or original, just real enough to force you to learn.
It’ll feel uncomfortable at first. Tools like the Terminal can be intimidating if you’ve never used them. But AI is there to explain everything and guide you through every step, and help you whenever you get stuck. You can work at your own pace with no pressure to ship.
What you’ll discover is that not only are you learning a lot, but there’s a huge sense of achievement from building something from nothing. It’s addictive. Developers have known this forever but designers and researchers might not have tasted it before.
As a designer or researcher, you already have the skills to think end-to-end about a product. You understand users, you understand the experience, you understand what good looks like. The only thing you’ve been missing is the technical bit at the end, and AI can close that gap. Go and get building and learning.
