Halo - Helping people stay in control of AI

What should people see before an AI takes action? How can they check its sources, follow its progress, or undo a mistake?

I created Halo to explore these questions through an open-source React library of 12 interactive components. It covers patterns such as human approval, source transparency, execution status and recovery.

I took the exploration beyond static screens into working components with documentation and light and dark themes. The project brings together my interests in interaction design, reusable systems and building with AI tools, while making deliberate decisions about where people need visibility and control.