Laurie Voss argues that coding agents reduce the cost of producing codeCode generation uses AI or another automated system to create source code from instructions, examples, schemas, or higher-level specifications. while review, operational work and security remain bottlenecks. He discusses gaps between passing automated tests and a maintainer deciding a change is ready to merge.
Voss predicts that software teams will put greater value on understanding users and describing what a successful product should do. His longer-term claims depend on the assumption that agents will improve at review, operations and security; he acknowledges that this remains a prediction.
Voss describes the difficulty of training junior developers when entry-level coding tasks are delegated to agentsAI agent delegation assigns a bounded task, context, authority, and expected result from one participant to an agent or another agent.. He suggests growing product-oriented skills deliberately, writing down expert judgmentHuman judgment in AI is the accountable interpretation and decision-making people contribute when setting goals, evaluating evidence, and managing consequences. as specifications and evaluationsEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria., and measuring delivered outcomes rather than code volume.
Watch on YouTube
