Chatlyn

Context

Chatlyn is a fast-growing Viennese SaaS company building an AI-powered communication hub for the hospitality industry. With a recent Series A and ambitions to scale globally, the product needed to grow up fast.

The challenge

When I joined, the Figma files didn’t reflect the live product and were essentially unusable as a design foundation. The codebase told the same story: the same components existed in multiple inconsistent variants with no shared logic between them. There was no design system, no single source of truth, and no way to move fast without making things worse.

What I did

I started by attempting to map the existing app one to one, but it quickly became clear that was a dead end. The inconsistency ran too deep. Instead I made the case for building a proper design system from the ground up, aligned the team around it, and got the engineering team on board with adopting PrimeVue as the component foundation. I set up the design system in Figma, built new components styled to match the product’s visual identity, and recreated the app within it. When the team later decided to move to Shadcn, the design system moved with it.

Beyond the system itself, I introduced design thinking and structured usability testing to the team. Research on the Guest CRM feature challenged assumptions about how hotel staff actually works, sharpened the product’s ICP and gave the team a much clearer picture of who they were designing for.

I also established AI as a core part of the design workflow. Where AI had barely been used before, it now works directly inside our tools via MCP integrations, understands our design system, and bridges the gap between design and development. Designers can write requirements and have AI draft directly in Figma using valid, system-compliant components. The result is a step change in speed, quality and consistency across both disciplines.

The trade-off

I first tried to mirror the live product one-to-one, the safe, incremental path that wouldn’t rock the boat. But the inconsistency ran too deep and it was a dead end, so I made the harder call: rebuild from a proper design system and move engineering onto a shared component foundation. The short-term cost of rebuilding bought a system that later survived a full framework migration without falling apart.

The result

A product team that went from Figma chaos to a scalable, AI-augmented design operation. Faster output, stronger consistency, and a shared foundation that design and development can both trust.