Relay — Supervision Interfaces for AI Agents
A supervision workspace for long-running enterprise agents, built so their work is observable, interruptible and recoverable.
Relay — Supervision Interfaces for AI Agents
A research prototype for supervising long-running AI agents. I designed a modular workspace where people can understand progress, trace sources and steer work across enterprise systems.
Early shared-workspace prototype
Problem
Delegation still meant juggling tools and disconnected context. People needed to see what agents were doing and know when to step in.
Results
Research shaped a single task view, guided setup and role-based Packs and Skills. The interface brings progress and points of control into one workspace.



















