This page reflects two directions I am actively shaping outside of my core academic profile: personal super intelligence and agentic AI infrastructure.

Personal Super Intelligence

Personal AI systems that become coherent, persistent instruments for one person's life.

I am interested in AI systems that do more than answer questions in chat. The longer-term direction is a personal computing environment where generated tools, interfaces, and agents share state over time and become usable instruments for reflection, planning, sensing, and day-to-day action.

This sits at the intersection of intent modeling, goal tracking, and AI-generated software. The goal is not a generic assistant, but a system that lives with you long enough to know what you're pursuing — one that turns scattered intentions into coherent goals, and coherent goals into a visible trail of things done in this AI age.

Check out our new arXiv!

PSI: Shared State as the Missing Layer for Coherent AI-Generated Instruments in Personal AI Agents

Zhiyuan Wang, Erzhen Hu, Mark Rucker, Laura Barnes

A shared-state architecture that turns independently generated AI modules into coherent instruments through a personal-context bus. Demonstrated across 14 modules in a three-week autobiographical deployment.

Agentic AI Infrastructure

The systems layer that makes AI agents reliable enough to operate as software.

I build infrastructure for long-running agents: orchestration, tool use, operator controls, review loops, and developer-facing workflows. My interest is in the connective tissue that lets agents interact with tools, memory, and each other in ways that remain observable and controllable.

This is a systems problem, not just a prompting problem. The hard part is keeping agents stateful, composable, and operational over time, with interfaces that real developers can actually use.

More coming soon...