Introduction

What happens when you collaborate with an AI to design its own interface? This design study documents a unique partnership with Claude, where we explored how to translate its self-understood traits into tangible design decisions. Instead of imposing human assumptions about AI interaction, I let Claude’s character assist in guiding the design process.

This exploration draws deeply from the groundbreaking work of Amanda Askell, a philosopher at Anthropic whose research focuses on AI alignment and character development. Her approach to training models for specific traits—particularly honesty and thoughtfulness—inspired a central question: How might we design an interface that authentically reflects these carefully developed characteristics?

Rather than treating interface design as a purely human endeavor, I engaged with Claude as a collaborative partner in reimagining its own user experience. This partnership allowed us to translate Claude’s self-perception and core traits into tangible design principles, creating an interface that genuinely embodies its character while honoring Anthropic’s commitment to developing truthful and thoughtful AI.