Designing for agentic AI requires new UX research methods focused on trust and accountability¶
Insight: Agentic AI — systems that plan, execute, and persist in tasks autonomously — requires UX research to move beyond usability testing into trust, consent, and accountability design. The key distinction from RPA: agentic AI formulates plans based on goals rather than following predefined scripts. RPA mimics human hands; agentic AI mimics human reasoning. UX teams need new research playbooks that address when systems make autonomous decisions on users' behalf.
Detail: Yocco distinguishes three levels: predictive AI (flags conflicts, user acts), generative AI (creates content, user reviews), and agentic AI (identifies problems, formulates solutions, takes action). The design challenge shifts from "did the user understand the output?" to "does the user trust the system to act on their behalf?" with implications for consent mechanisms, transparency of reasoning, and accountability when agents make mistakes.