A calmer way to build products with AI.
The starting problem was not generation speed. It was control. AI builders can help people produce an interface quickly, but the surrounding workflow can feel fragmented and difficult to trust. I designed Aurora to connect prompting, project management, live preview, editable code and credible sources in one coherent product experience.
- Role
- Product designer
- Contribution
- Strategy, UX/UI, visual direction, design system
01 / Project foundation
Make the next action clear before asking users to build.
A generative workspace still needs the fundamentals of a dependable product. I designed flexible grid and list views, purposeful empty states and clear project entry points so users can understand what exists, resume work quickly and start without facing an intimidating blank canvas.
02 / Building with control
Keep the speed of prompting without hiding the work.
I treated the prompt as the beginning of the workflow, not the entire product. References can be attached directly, generated output can be reviewed in a live preview and the underlying code remains accessible. This gives less-technical users a clear path forward while preserving control for people who need to inspect or refine the result.
03 / Trust and transparency
Let people verify an AI answer without breaking their flow.
Generated answers are more useful when their evidence is easy to inspect. I designed citations as part of the response itself, with lightweight source previews and a deeper source view when needed. The interaction keeps verification close to the answer instead of sending users into a separate research process.
04 / Community and access
Turn finished work into a useful starting point for others.
The community layer extends Aurora beyond private creation. People can discover how others approached a problem, understand the result and use shared projects as inspiration. Pricing was designed alongside this journey so the path from exploration to choosing a plan remains clear and consistent.
05 / Product system
Build consistency into the product, not just the final screens.
I developed reusable interface patterns for chat, navigation, inputs, actions and feedback across light and dark environments. Exploring early directions alongside the component system helped establish a visual language that could support new workflows without each feature feeling like a separate product.
The result
A product direction that balances AI speed with structure, control and trust.
Aurora brings the full building journey into one system: starting and organising projects, generating from prompts and references, reviewing the interface, inspecting code and validating source material. As an independent concept, it demonstrates the product principles and interaction model rather than claiming untested business impact.