Fitz Zhang

Engineer, Personal Agent Bench

Builds the v2 record-and-replay exam on Personal Agent Bench. Previously a frontend engineer at ByteDance and Tencent Docs.

medium.com/@591763389 · x.com/zcl_2014

Posts

• Record and replay (v2)How to test AI agents on real websites: record a site once, replay it offline, and every run of the exam sees the same pages, prices and clocks.