Designing in Public

Hi! I’m Alli, a design/engineer hybrid. I recently stepped down from my role as Head of UI/UX at the startup I cofounded. I’ve worked in startup engineering for over a decade, and now I’m working on something new:

Behold, Designinator!

Also, I’m a big Phineas and Ferb fan.

Why Designinator?

I’ve spent most of my career in startups, and I wouldn’t have it any other way. I’ve worked on multiple early-stage products, where speed matters more than polish. The UI starts as “just Tailwind classes” and “shared components” with the idea that “we’ll clean it up later.”

When “later” arrives, it hurts:

• For some reason there’s three “dialog” components, and they’re all being used. Meanwhile,  divs that should be shared components have been copied and pasted all over the place.

• Divs are styled up as buttons and made to look clickable, without any consideration of accessibility.

• No one is using the “secondary” color; instead, Tailwind’s teal-500 is all over the place. Plus, there’s 50 shades of gray in one application.

• Your AI coding tool is inventing text colors and pairing them with backgrounds that are low-contrast and unreadable.

I’ve been there. I tried prompting my way out of the mess, and while it helped, it still took real time and effort to shape a design system I felt comfortable both using myself and handing off to AI as part of my development process.

Why design systems matter early

Having a defined color palette, reusable components, and a shared language helps engineers (both human and AI) quickly prototype ideas and ship new features without a bunch of clean-up after. Design systems are even more important when you’re building components you can’t just pull straight from ShadCN.

Without a design system, velocity looks fast but slows over time.

When you go back and try to retrofit a design system, you find yourself dealing with one-offs and outliers, inconsistent behaviors, and CSS that many engineers are afraid to touch. You’re fighting entropy, not designing.

Enter Designinator

Designinator is an opinionated tool for surfacing existing design patterns and organizing what already exists. It focuses on class sprawl by analyzing usage across a codebase and helps identify where “shared” isn’t actually shared. It’s not a magic button or a design system in a box. It’s not intended as a replacement, but as a tool for designers and engineers that works alongside existing frameworks.

Why No AI (Yet)

If you look through the Designinator source code, you might notice that it doesn’t use AI. This is intentional. To start, I want Designinator to have deterministic outputs, clear reasoning, and predictable behavior. This way, it’s clear exactly what Designinator is doing. In the future, AI can function as an eventual layer on top of understanding, not a substitute.

Why Open Source

Design systems are deeply contextual, and when it comes to codebases, trust is important. Keeping Designinator open source allows me to show how decisions are made, get feedback from the community, and enable users to adapt it for their needs.

I’ve learned more from using open source tools than closed ones, and I want to contribute back to the community that’s taught me so much. I also have personal reasons for making it open source. I’m building in public, and for this project that means showing my work. I’m more interested in learning while shipping than rushing a product to market. I’m not trying to corner the market. I’m learning how to build an open source product.

Building in Public

I’m really excited about building this tool in public, so I can share my thought process around my design decisions, tradeoffs, what works (or doesn’t). I’ve done most of my design and engineering work in private repos and behind NDAs, and I’m excited to share what I’m working on.

This is a side project. I’m not going to be posting daily updates or promising features that don’t exist. Instead, I’m here to have a conversation and learn alongside others. If you’ve felt the pain of creating a design system after building a product, tell me how you handled it.