Nate B Jones gives both agents access to the same problem-discovery skillAn AI agent skill is a reusable package of instructions, resources, and tool guidance for performing a bounded kind of work. and asks them to inspect a working repository. The results reveal different priorities: Fable focuses on product and workflow pain, while Codex gravitates toward defects it can concretely verify and fix.
Nate B Jones argues that the comparison exposes a hidden source of agent quality. Two capable systems can produce useful work while optimizing for different definitions of importance, so asking several agents to diagnose the same environment can uncover a broader opportunity set.
Nate B Jones recommends matching the model to the stage of workAI model routing sends each request to a model chosen for that request's complexity, cost, speed, privacy, or other requirements.. Exploratory systems can frame and rank problemsProblem framing with AI is the human-led process of defining the real goal, constraints, evidence and success criteria before using a model to explore solutions., then clearly scoped implementation tasks can move to agentsAn AI coding agent is a tool-using AI system that can inspect, modify, and validate software within a repository. that are dependable at finishing them.
Watch on YouTube




