Core topics
Trust and evidence, human control, agent visibility, review before action, progressive disclosure, uncertainty, recovery, memory, and privacy.
Use topics to connect patterns
A topic is broader than a single component. For example, human oversight may involve previews, approvals, activity logs, undo, and escalation together.
Open the full atlas to see which patterns support each topic and where their trade-offs differ.
Explore topics in AI Interaction Atlas
Categories · Patterns · Compare · Search