A record of all updates since the course opened in August 2026. All of these updates are free of charge for existing students.
Five new lessons. “Mental models” covers three fundamental principles that the course is based on. “What are skills” explains what a skill actually is, and how it can influence your agents, in the right way.
“Get good at telling AI what you want” is about how the right word influences the agent’s output. Lots of interesting examples in this one.
the share menu feels janky, make it smoother
transform-origin: center; transition: all 400ms ease-in-out;Switch between the two prompts to see the difference.
“Developing taste” explains why taste is the differentiator now, and how to train it.
“Knowing where to learn from” is about choosing who and what is actually worth learning from.
There’s also a new vocabulary page with 244 design and animation terms across 22 categories, each with an interactive demo. Terms are linked inline in lessons too, so you can hover a word to see what it means.
The timer needs tabular-nums, the labels need more tracking.
A new lesson on a workflow I use often: turning everything someone has shared publicly into a skill, so you can ask them questions while you work. It comes with the /emil-borrow-a-brain skill that does the heavy lifting.
/emil-borrow-a-brain build a skill from everything Paul Graham has written at paulgraham.com
The second lesson, “How to build with aiforui skills”, explains which skill to run, when, and in what order. We take a generated settings page and run it through the skills one by one until it’s ready to ship.
Same screen, before and after /emil-design-foundations.
/emil-product-thinking – Helps you come up with the right idea, which matters more than ever now./emil-prep-for-prod – Preps your app or branch for production. One of my most used skills./emil-borrow-a-brain – Turns someone’s writing into a skill you can ask questions./emil-mobile-native – Makes your app feel better on mobile.A new workflow with two lessons on how I use AI day to day. “Using AI at work” covers what I hand off to an agent at my day job and in my personal projects, and what I keep for myself. You can test your own judgement on it too.
A ⌘K command menu
Fuzzy search, keyboard navigation, nested pages, recent items. You know the pattern, and so does the agent.
“Defining a system” is about turning your design system into something an agent can read. When your tokens are named and documented properly, the agent can start using them properly.
The variables don’t change, only the values they point at.
There’s also a new lesson in the fundamentals section called “Friction as a feature”. It’s about why some friction is a natural filter for bad ideas.
All skills are now prefixed with emil- so they are easier to find.
/emil-unslop-code – Removes verbose comments and other AI leftovers. I run it before every PR./emil-unslop-design – Removes AI slop from your designs./emil-unslop-writing – Removes the AI slop that gets added to text./emil-design-system-docs – Builds design system docs following the rules from “Defining a system”.A new workflow with three lessons on how I use AI to build tools that help me build things. Some of them are practical, like a border radius inspector. Others are more creative, like generating new interactions and patterns.
One of them is a poster press. Prompting for posters directly gets you stuck in a loop of small corrections, so instead we build a tool and use it to explore directions we wouldn’t have thought to ask for.
The last one is about review. Instead of squinting at nested border radii, we build an inspector that checks them for us.
Press inspect and hover over the two cards.
/emil-get-creative – Helps you explore and find the right direction./emil-build-a-tool – Helps you break out of the prompting loop.A new lesson called “Prototyping interfaces”. It explains why prototyping is essential and how to do it well with AI. Instead of asking for one version and fixing it, you ask for several and pick the best one.
We’ll use this to create your course account.
There’s also a new workflow called “Reviewing animations”. It’s built around Lapse, a tool I recently acquired from Matt. It records an interaction frame by frame and hands it to your agent, so the agent can fix an animation instead of guessing at it. Lapse is available to course members only for now.
/emil-prototype – Builds multiple versions of a UI piece behind a switcher./emil-ask-emil – A router over every skill in the course./emil-ask-lapse – The manual for Lapse.
Join the waitlist to get two free lessons and be the first to know when enrollment reopens.
Enrollment reopens on Tuesday, December 8