Riley Brown discusses his experience comparing successive Claude Opus releases and cites a coding evaluation that weighs performance against time and costEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria.. His impressions and the benchmark results are presented as observations, not universal rankings.
Riley Brown demonstrates desktop agent features including computer use, browser sessions, linked applicationsAgent tool use lets an agent call bounded external capabilities so it can retrieve information or change state beyond generating a response., chat creation and search. He considers how these capabilities might change the workflow for building software.
Riley Brown proposes agent-native mini applications that present tailored interfaces connected to existing tools, using email review as an example. He describes this as a desired future capability rather than a feature he has established is already available.
Watch on YouTube




