Precious Plastic is an open-source platform enabling people worldwide to recycle plastic locally — building machines, running community workshops, and sharing knowledge freely. The mission is compelling, but the platform's depth was working against it. Hundreds of guides, forum threads, machine manuals, and community map tools existed, but discovering the right resource required users to already know where to look.
Two distinct user types were failing in different ways. Newcomers — home recyclers trying to set up their first machine safely — were getting lost in technical documentation written for makers, not beginners. Community organisers were struggling to surface local spaces and events buried across disconnected tools. High intent, low follow-through — and in the case of machine setup, a real safety risk.
Solo UX/UI designer, end-to-end ownership across research, information architecture, user flows, and high-fidelity prototype. Three-week sprint on responsive web. The brief I set myself: design an AI assistant layer that reduces time-to-answer for both personas without rebuilding the underlying platform — an integration challenge as much as a design one.
Before vs After: Manual navigation, information overload, limited guidance → AI-powered voice/chat assistant, personalised support, seamless resource discovery.
Research first. Hypothetical survey of 100 users (20–45, urban DIY recyclers and sustainability advocates) to establish a problem hierarchy before designing anything:
The pattern was clear: the content users needed existed — the platform just had no way to meet them where they were.
Mapping the current failure. An IA audit revealed that the existing flow was entirely linear — separate tools for each task, menu-driven navigation, no personalisation. Users had to know the right section before they could find the right answer. The redesigned flow needed to invert this: intent-first, with the assistant routing users to information rather than expecting them to navigate to it.