Borys Kosmynka AI coding assistants can now write working Python in minutes – but for type designers, “working” and “correct” aren’t the same thing. This workshop is a hands-on taster in using AI as a collaborator for building your own tools, not a coding class: after a short orientation to the toolkit (RoboFont’s scripting console, a code editor, the terminal, version control) and just enough Python, JSON and XML literacy to recognize what you’re looking at, most of the session is spent experimenting directly – framing a small idea from your own work, prompting for it, and reading the result critically. The focus is the workflow itself: knowing when to let go, and when to try and do some of the block building ourselves. Understanding the gaps between errors, code that runs and code that’s correct. Regular group check-ins throughout surface what’s actually going wrong in the room, turning real mistakes into shared lessons.
No coding background required or assumed – by the end you’ll have a working method for testing an idea before trusting it, and a realistic sense of where AI helps and where it quietly misleads.
Each participant needs:
- A GitHub account
- A laptop with RoboFont installed (licensed or the free/demo version – flagged in advance so nobody arrives stuck on installation)
- VS Code – open source project code editor
- Sublime Text – single file code editor
- Terminal access (built in on macOS/Linux; WSL or Git Bash on Windows)
- An active subscription to an AI coding assistant (Claude Pro or ChatGPT Plus/Team) – participants should check their own weekly usage isn’t already near its cap before the day
- A UFO of their own to work from – required, not optional