First I want you to think of a few categories of your UX knowledge…how you would read a project brief, begin to craft user research, perform some competitive analysis, prototype, check for accessibility – well, you get it. Outline your process in your mind. Now remind yourself that most of what you have been working with regarding AI are LLMs….Large Language Models. They have retained what they have scraped from the internet and a billion books, but what good is that to you? When you ask it a question, it will look into its pile of obtained knowledge and retrieve for you a prioritized result. Not personalized. Not structured for the way you work. Not organized using the steps your Agile or SAFe team works. And that is the heart of the matter today. Getting the most for your tokens out of AI, by sharing your UX work methods with it to obtain optimal results.

I’m not getting into the nitty gritty of teaching Github here, but I will link to some code, and illustrate how to stop foisting a singular (albeit perhaps compound) command into AI and getting out a singular result. That’s one part of my approach. This is more regarding automation, and how your can teach AI some of what you know about UX, to then get back out of it a valuable result. The other part of this is that I know your recognise, you’re the only one who knows when to run a heuristic eval instead of a usability test. An even more specific use case would be, for a larger task like auditing a design system, where you need it to work through token usage, naming conventions, a11y and theming support – and you don’t want to tell the AI to do this each time.

For those of you who are Github savvy, the repository is located here. You can browse the code and read the plain text commands to see how things are structured. Code is under the MIT license, so feel free to fork it, use it or just browse for ideas. As far as usage, let me give you some examples:

1. Run a research discovery cycle type /design-research:discover then tell Claude: “I’m designing a habit-tracking app for people recovering from burnout. Run a full discovery cycle – personas, empathy map, and journey map.”

2. Generate an accessible color palette type /ui-design:color-palette: then tell Claude: “Create a color system for a health and wellness brand. I need primary, secondary, neutral, and semantic colors, all passing WCAG AA standards.”

3. Build a developer handoff package type /design-ops:handoff: then tell Claude: “Generate a handoff spec for a settings page with a profile section, notification preferences, and a danger zone for account deletion.”

4. Run a heuristic evaluation type /prototyping-testing:evaluate: then tell Claude: “Run a heuristic evaluation of a checkout flow with these steps: cart review, shipping address, payment, and confirmation. Flag usability issues by severity.”

5. Write a case study for your portfolio type /designer-toolkit: write-case-study: then tell Claude: “Help me write a portfolio case study. The project was a redesign of an internal dashboard for a logistics company. We reduced task completion time by 45%. Walk me through the full structure.”

Each one gives you something tangible in a few minutes: a full persona set, a production-ready color system or a handoff doc that engineers can actually use.

I’m curious how you use tools like this. Do you like Claude Code…does your company even use the same AI, or do you have different ideas on how to personalize AI for UX? Feel free to drop me a line at brent.rish@gmail.com.