We’re Changing Up Dev Drop…But Can It Run Doom?
Hello everyone! Welcome to the new and improved Dev Drop. We are going to tighten up our list of links, aiming to share the very best of what we are publishing on Builder and finding around the web. We’ll also be publishing more frequently, with the goal of ensuring that the links we share with you aren’t stale by the time they hit your inbox. This week we have an interactive guide to automating your design-to-code workflow, AI music getting scary good, and Doom running as…Typescript types?
Figma-to-Code - Why Do The Same Work Twice?
Check out our pragmatic guide for engineering leaders looking to measure the impact design-to-code automation could have on their teams. We’ve got examples and instructions on how to make it happen with your design system.
How to optimize a Figma design to ship AI-powered experiences
Our upcoming Builder Lab livestream will focus on how to optimize a Figma design to get the best results during an AI-powered import to Builder.
I’m Too Young To Die
Someone remade DOOM... using only TypeScript types. The project is 3.5 trillion lines of types, takes 90GB of RAM to process, and compiles one frame every 12 days, but hey, at least we've proved "for sure, for sure" that TypeScript types are Turing Complete.???
A New Wave Blows In
We recently ranked our top AI-powered IDEs and Windsurf was among our favorites. Windsurf Wave 4?dropped this week, and it finally solves those context-switching headaches, placing text filters, file attachments, API requests, and HTML analysis right in your editor.
Fine Tuning Music Models
Music made by artificial intelligence took a big step forward in February, to the point where we might call it scary good. A recent paper detailed the work of NotaGen, a new symbolic music generation model trained on multiple periods of classical music and even pop. The resulting scores and related sheet music are surprisingly diverse and sound pretty darn good.
A Design Centric LLM
Researchers published an interesting paper on Brickify, a “visual-centric interaction paradigm to specify design intent for controllable image generation” The concept is that users start with reference images and identify design elements such as subjects, styles, colors, and concepts. These design elements can be made into interactive, reusable design tokens. User can then then directly interact with these tokens to build a visual lexicon and instruct the model on how to combine them to craft larger holistic imagery.
Hey, you’re already at the end. If you have opinions on what you would like to see in Dev Drop or how you feel about the new format, reach out and let us know - [email protected]