Speak the Domain: How DSLs Help Humans — and AI — Model the Real World

What if instead of writing code to describe something, you could write a language about it? That's the idea behind domain-specific languages, and once you've built one, you'll never stop seeing where they belong.

In this talk, Jake introduces DSLs from first principles: what they are, why they're worth reaching for, and how they let you model messy real-world problems with surprising clarity. He'll walk through Tamarack, an open DSL he's building for video and image annotation, as a hands-on case study — showing how a well-designed language can describe scenes, sync media, and track annotations in a way that's human-readable, git-friendly, and built from the ground up to be understood by AI.

That last part matters more than ever. As AI becomes a collaborator in our workflows, the structure and intentionality of a DSL gives it something powerful: a constrained, meaningful vocabulary to work with — making its output more predictable, more trustworthy, and more useful. You'll also get a practical, beginner-friendly look at how to actually build one — from identifying the right domain to designing a grammar your team can read on day one. You'll leave with a new mental model and the tools to start speaking your own domain.

• Development
• Emerging Tech
• AI/ML