Reflek.io

The design process focused on translating a highly abstract system into an interface that felt structured, logical, and usable. I mapped end-to-end user flows covering ontology creation, twin definition, data integration, visualisation, and real-world control, using wireframes and flow diagrams to validate mental models before moving into high-fidelity design. The UI was built in Figma using a modified Mantine design system, which I treated as a core product deliverable rather than a supporting layer. Components were adapted, extended, and created to handle dense data relationships, complex configuration states, and AI-assisted workflows, ensuring consistency and scalability for a small, distributed engineering team. A key part of the experience was designing for real-world interaction and AI integration. Users could connect sensors, define behaviours, and trigger actions controlling physical systems such as robotic arms or environmental conditions. Alongside this, Airwin - the platform’s built-in AI and LLM capability, was designed into the product from the outset. In the MVP, Airwin provided a chat-based interface that helped users understand, model, and evolve the ontology through conversation, requiring careful attention to transparency, feedback, and user control to ensure trust in AI-assisted decisions.