# Articles

Notes on design, side projects and life in between.

- [What a team loses when designers ship their own code](https://bartdunweg.com/articles/what-breaks-when-the-designer-ships) (Oct 13, 2026): Designers can now take their own work live, and the speed is real. So is the loss: a second reader, an engineer's objections, and code that more than one person owns.
- [Where the design process goes when you work with AI](https://bartdunweg.com/articles/the-process-moved-into-the-setup) (Sep 8, 2026): The design process did not die when agents arrived. It left the meetings and moved into the files an agent reads, and a team that cannot see it has not dropped it.
- [How to define what a good AI answer looks like](https://bartdunweg.com/articles/evals-are-a-design-deliverable) (Aug 4, 2026): Every AI feature needs someone to write down what good output looks like. That is a design review made explicit and repeatable, and designers should be the ones writing it.
- [AI design tools keep the result and forget the reasons](https://bartdunweg.com/articles/the-tools-are-built-around-the-artefact) (Jun 30, 2026): AI design and coding tools give the output a whole interface and give the reasoning behind it none. So every generation keeps the screen and throws away why it looks that way.
- [What I tell an AI before it touches my design](https://bartdunweg.com/articles/the-standards-i-give-my-coding-agents) (May 26, 2026): An AI agent builds fast and forgets every design decision overnight. How I write down taste, rules and feedback so the next session builds inside the system.
- [How to judge what AI makes for you](https://bartdunweg.com/articles/the-judgment-layer) (Jan 6, 2026): An AI model will give you twelve competent options and an opinion about none of them. How I form my own opinion before I generate anything, and the three questions I ask before I accept what comes back.
- [When animation helps, and when it gets in the way](https://bartdunweg.com/articles/motion-that-survives-a-second-look) (Dec 2, 2025): The test of an animation is not whether it impresses once. It is whether it still feels right the fortieth time, which is a different problem entirely.
- [Performance is a design decision](https://bartdunweg.com/articles/performance-is-a-design-decision) (Jul 15, 2025): Speed is treated as an engineering concern, so designers hand over work that cannot be fast. Most of what makes a page feel slow was decided in the design file.
- [Form validation is a design problem](https://bartdunweg.com/articles/form-validation-is-a-design-problem) (Jun 10, 2025): Validation is the moment a product tells someone they are wrong. Almost every rule about it comes down to when you say it, and whether you say what to do next.
- [Design tokens that survive a rebrand](https://bartdunweg.com/articles/design-tokens-that-survive-a-rebrand) (Apr 1, 2025): Some products can change their brand colour in a day; others need months. The difference usually comes from how the design tokens, the named values a product's colours come from, were set up years earlier.
- [How an idea becomes a digital product, step by step](https://bartdunweg.com/articles/from-idea-to-product) (Feb 25, 2025): The steps between an idea and a working digital product, in the order they happen: what each one is for, what you should have in hand when it ends, and the common mistakes.
- [Why a design file is not a handoff](https://bartdunweg.com/articles/the-handoff-is-not-a-document) (Jan 21, 2025): Design handoff, the moment a design passes from designer to developer, keeps getting better tools. What goes missing is the reasoning behind the design, and a more detailed file does not carry that.
- [Design sprints that actually ship](https://bartdunweg.com/articles/design-sprints-that-actually-ship) (Dec 17, 2024): A design sprint is easy to run and hard to make count. Whether it changes what a company builds is decided before the week starts and after it ends, not during it.
- [Why most design systems quietly fail](https://bartdunweg.com/articles/on-design-systems) (Nov 12, 2024): Most design systems fail not for lack of components but because teams stop trusting them. It rarely gets said out loud: one team after another simply copies a button and changes it for themselves.

Every article is also Markdown at its own address plus `.md`, and the list is in the [RSS feed](https://bartdunweg.com/feed.xml).