VUX (Verification User Experience)
Coined by Sasha Merzliakov · First published April 2026
The design activities for interfacing with and managing AI agent teams, where the human role has shifted from tracking work in progress to verifying whether completed output is correct.
What does VUX mean?
VUX stands for Verification User Experience. It names the emerging design discipline that addresses what happens when the primary bottleneck in a workflow shifts from generating work to verifying it. When AI agents can produce high volumes of output faster than humans can review it, the human role changes: from coordinator to checker, from sprint manager to quality gate.
Every era of work management created interfaces shaped by its bottleneck. Waterfall needed Gantt charts to track the plan. Agile needed sprint boards to track priority. Kanban needed WIP columns to track throughput. hcd.ai argues that agent orchestration needs a new interface class entirely, one designed not for "is it done?" but for "is it right?"
Why VUX matters
The existing tools practitioners reach for — Jira, Linear, Notion, kanban boards — were designed to answer one question: is this work in progress, blocked, or done? That question becomes the wrong one when the production bottleneck is removed. AI agents do not wait. They complete tasks, return output, and queue the next one. Verification becomes the constraint.
The cognitive and interface challenges this creates are distinct from prior work management problems:
- Approval queues — surfacing which outputs need human review, in what order, ranked by what needs you next
- Compounding errors — small mistakes at step one propagate through every step that follows, per Lusser's Law
- Judgment calls — the human role is not accepting or rejecting binary outputs but exercising contextual expertise on borderline cases
- Input quality as lever — vague briefs amplify downstream variance; better input reduces verification load more than better tooling does
VUX in the Agentic AI literature
The term was coined by Sasha Merzliakov to fill a gap in the practitioner vocabulary for agentic AI. Related concepts include human-in-the-loop (HITL), agent orchestration, and approval queue design — but none of these named the specific experience of the person whose primary role has become verifying AI output at scale.
See the full article for patterns, orchestration stages, and practical design recommendations: VUX: The Verification User Experience for AI Agent Teams.