Building a Knowledge-Work Harness: From Coding Agents to Executable Strategy

Talk overview

Building a knowledge-work harness requires addressing the verification gap, as decisions lack the automated checks found in coding. This talk shares insights from a year of running an open-source, local-first agent system, focusing on components that survive the transition and the role of strategy as a protocol. A live demo illustrates an agent making a gated decision based on plain-text rules.