Brandon Thomason

Independent researcher working on governance and control for autonomous coding agents. I build the systems I study — production AI automation running live against a commercial API, and a multi-agent orchestration system with an independent verification layer. Current work is a pre-registered equivalence test of whether guardrail phrasing changes how often an agent violates it. Method commitment: every design decision grounded against published literature, every unsourced claim treated as unverified until confirmed at source.

Personal Website

https://osf.io/a8yf6

Projects

No grants recorded.