How Moonchild Became the Shared AI Canvas for Poplin's Product Team

Poplin has one product designer. She supports three pods, a consumer app, a Laundry Pro app, and a roadmap that doesn't wait. For a single-designer team serving two apps across 500+ cities, this wasn't a process problem — it was a math problem.

For most companies, the equation ends one way: hire more designers. Poplin took a different route. Instead of growing the design team to match the pace of product, they brought their design system into Moonchild and opened the design process up to the people already writing the specs and shipping the code.

The result isn't that Talia, Lead Product Designer at Poplin, does less design. It's that she does more of the design that actually requires a designer — and less of the translation work that used to sit between a PRD and a Figma file.

The bottleneck wasn't Talia. It was the shape of the workflow.

At Poplin, every product idea used to pass through the same desk. A PM at Poplin like Karina Orton, would write a PRD, hand it to Talia, wait for mockups, review, iterate, wait again. Engineering would pick up the work a cycle later.

The instinct in most product orgs is to protect the designer's time by giving them more designers. Poplin's read was subtler: the designer's time was being spent in the wrong places. Talia's craft wasn't the bottleneck. The translation between product intent and design exploration was. And translation is exactly the kind of work that doesn't need a designer's hands — as long as whatever comes out the other end respects the system.

That "as long as" is doing a lot of work. It's also where consistency and maintaining a design system comes in.

The design system as the unlock

Moonchild is built for exactly this moment: turning your design system into a working environment where product, design, and engineering can all contribute to building product without breaking consistency. Every screen, flow, and prototype is generated on top of your actual system and can move straight into code.

Before Moonchild could be useful to Poplin, it had to know Poplin. Moonchild recreated Poplin's design system apples-to-apples — the bespoke monospace typography, the component library, the brand tokens, the patterns Talia had spent months codifying.

This wasn't a nice-to-have. It's the entire reason the rest of the story works. A PM generating screens on a generic AI tool produces something that looks like design but ships like noise — off-system components, wrong type, patterns a designer will have to rebuild from scratch for every new tool they try. A PM generating screens on top of Poplin's actual DS produces something Talia can review, refine, and send forward without rebuilding. As Talia puts it, Moonchild has "improved collaboration with product partners by helping them create designs that align more closely with the Design System from the start."

Talia's role shifted accordingly. She's no longer the person who draws every screen. She's the person who owns the system that lets every screen get drawn. That's a higher-leverage job, and it's one a single designer can actually do.

Because Poplin is on Moonchild's Team plan, the Moonchild team worked directly with Talia to set up Poplin's design system from their Figma and GitHub — porting over components, styles, tokens, and the custom guidelines, nuances, and rules that make the system Poplin's. A Moonchild DS isn't a static import; it's a living document. Poplin iterates on it inside Moonchild, uses it to generate UI, and exports it into wherever the engineering work happens — Claude, Cursor, Codex.

What changed for the PMs

Karina recalls her first in-depth Moonchild session working through a PRD for a shoe cleaning service — a pressure test for how Poplin will template other verticals like rug cleaning, ironing, and car seat cleaning. She wasn't blocked once.

More interestingly, the tool changed what she was doing. She started out uploading a PRD and ended up discovering gaps in her own spec: oh, I need to account for this and that. The act of seeing the flow on a canvas, in Poplin's actual components, surfaced product decisions that a text document would have carried all the way to design review before anyone noticed.

"Product jamming, getting to clarity with myself, and then streamlining alignment discussions with design, eng, and stakeholders." — Karina Orton, Product Manager, Poplin

This is the quiet shift. Karina isn't designing in the craft sense — that's still Talia's job. But she's arriving at roadmap conversations and design reviews with screens that already exist, in the right components, with the flow articulated. As Karina describes it, she's giving Talia "a launching off point and the context for her to at least know where my head is before she does the more refined UX work." The conversation is now about refinement, not translation.

That shift shows up in design review itself. Talia notes that "instead of spending time explaining that certain components don't exist or need to be redesigned, we can focus on the feasibility and quality of the solution itself." Reviews are productive earlier because everyone is reacting to the same on-system artifact.

The Poplin team has also felt the ship cadence. When they've flagged gaps or suggested improvements, those requests have turned into shipped features in days, not quarters. A feature Andrew Koch, Product Manager at Poplin, raised on a Wednesday was live the following week — prompting a "now that's a feedback loop" from him in Slack. It's the kind of responsiveness that makes a team comfortable going deeper on a tool, not just wider.

What it means for engineering

Moonchild's output isn't pixels. It's on-system screens that export as code, with MCP connections into the tools the team already uses. For Eric, Software Engineer at Poplin, and Azra Mirza, Engineering Manager at Poplin, that changes what "handoff" means.

"Velocity improved because engineers are spending more time building and less time decoding intent." — Azra Mirza, Engineering Manager, Poplin

The clearest test case is the Laundry Pro app. LP is currently on an Angular codebase that doesn't conform to the design system, and Poplin is migrating it to React Native. The plan: extract the business logic from the existing app, apply the DS from Moonchild, and rebuild with AI agents doing a significant share of the work. The Laundry Pro DS will share components with the consumer DS where it makes sense so improvements can ship to both apps at once.

This is where the story goes beyond design. The design system isn't just ensuring visual consistency. It's becoming the layer that allows product decisions, design outputs, and engineering implementation to stay aligned even as they move faster and more independently.

The New Default

It feels less like another AI tool, and more like infrastructure for modern product development. That's how Azra describes Moonchild's role at Poplin, and it's a useful frame for what's actually happening here.

Most teams facing this problem hire more designers. Poplin didn't. They made it possible for the rest of the team to participate in design — without lowering the bar.

That only works if two things are true: the system is accurate enough to trust, and the outputs are structured enough to build from. Moonchild sits between those two. It gives PMs and engineers a way to contribute to design inside the system, not around it — and it gives designers a way to scale their impact without becoming a bottleneck.

The interesting question isn't whether AI can generate UI anymore. It's whether that UI can be used. For customers like Poplin, the answer is increasingly yes — not because AI got better in isolation, but because it's grounded in a real design system and a real workflow. That changes who gets to contribute. It changes when decisions get made. And it removes an entire layer of translation that used to slow everything down.

What's Next

Poplin is extending how it uses Moonchild: expanding the design system to cover more real-world states and patterns, bringing engineers directly into iteration loops, and moving closer to a true PRD → system → code workflow.

This is the workflow Moonchild is built to power — AI-generated UI grounded in a real design system, usable by the whole product team, and exportable to code. Poplin is one of a growing set of teams using it to scale their design function without scaling their design team.