zain_
ai lab · intervention
← all of the ai labProduct2026industry

Unjargon: Dev-Speak, Translated for Designers

Designers work next to engineers all day and quietly drown in vocabulary nobody stops to explain: props, endpoints, CI, tokens, hydration. Unjargon is a live glossary that translates dev terminology into plain language and everyday analogies, with an AI that can generate a fresh explanation for any term it doesn't have yet.

role
Solo build: product, design, and development
stack
Next.js · React · Supabase · Vercel AI SDK · Gemini · Tailwind · shadcn/ui · TypeScript
status
shipped · in use

How I got here

I've spent years as the designer sitting next to engineers, and I noticed something I never see written down: designers absorb an enormous amount of dev vocabulary by osmosis, and most of it is never actually explained. You nod along to "just pass it as a prop," "it's a hydration issue," "the token isn't mapped," and you half-guess the meaning from context for months.

I didn't want a dictionary that explained jargon with more jargon. I wanted the version I'd have wanted on day one: the plain-English meaning, and an analogy that makes it stick.

How I thought about it

Two rules shaped it.

Analogy first. A definition tells you what a word means; an analogy tells you where it lives in your head. Every entry leads with the everyday image, then the precise explanation underneath.

It can't be a fixed list. Dev vocabulary is effectively infinite and always growing, and a glossary that only knows what I pre-loaded would be stale in a week. So the product had to be able to explain a term it had never seen, on demand, and keep it.

I organized the whole thing the way a designer actually meets these words: by territory, not alphabetically: core languages, styling, building blocks, the backend, the workspace, package managers, version control, launching, documentation. Ten shelves you can browse instead of one flat A–Z wall.

What I actually did

I designed and built the whole thing myself: product, UI, and code.

The core is a searchable glossary backed by a real database, browsable by those ten categories, with instant search across every term. There are two ways to read it: a dictionary mode for scanning and looking things up, and a flashcard mode that flips term-to-explanation for when you actually want to learn the set rather than reference it.

The part I'm proudest of is the AI generate flow: when a term isn't in the glossary, you can ask for it, and an AI model writes a fresh explanation in the house style (plain-language description plus an analogy) which then gets reviewed and saved into the right category. The glossary grows itself instead of going stale.

It's a proper product, not a toy: light and dark themes, keyboard-first search, responsive down to the phone, and an admin path for curating what gets added so quality stays high.

photo drops here
The glossary in dictionary mode: a dev term (e.g. 'hydration' or 'prop') shown with its plain-language explanation and the everyday analogy underneath, category rail on the left
photo drops here
Flashcard mode mid-flip: the term on the front, its analogy-first explanation revealed on the back, for learning a category rather than just looking one up
photo drops here
The AI generate flow: typing in a term the glossary doesn't have yet, and the model returning a house-style description + analogy ready to review and save

Where it landed

Unjargon is built and deployed, a working public product on the modern stack I actually reach for: Next.js and React on the front, Supabase for the glossary store, and the Vercel AI SDK wired to Gemini for the generate flow.

It came out of a specific belief I keep coming back to: the gap between design and engineering is usually a language gap before it's a skill gap. Close the vocabulary, and the collaboration gets noticeably easier. Unjargon is my small, shipped attempt to close it for the next designer who's quietly guessing.