Gergely shares insights on a growing trend among software engineers: using multiple AI agents simultaneously to boost productivity. With tools like Claude Code and OpenAI Codex becoming widely available, engineers are experimenting with running several agents at once, often on different tasks. Sid Bidasaria from Anthropic mentioned that using these agents improved his productivity. Simon Willison, a recognized expert in AI engineering, initially skeptical, now embraces this approach for tasks that don't require constant oversight, suggesting it could change how engineers work. The article highlights how traditional software engineering relies on a flow state—understanding components, building solutions, and submitting code for review. This process can be disrupted by interruptions, which are common in the field. Senior engineers, who often juggle multiple tasks and manage interruptions well, may adapt more easily to using parallel agents. However, not all engineers find this approach effective. Armin Ronacher, creator of Flask, expressed that while he occasionally uses parallel agents, the cognitive load can be overwhelming. Gergely shares his personal experience with AI agents for side projects, emphasizing the importance of testing, task clarity, and regular code reviews. He notes that setting practices, like ensuring AI-generated code passes all tests before proceeding, leads to better outcomes. The article suggests that while there’s potential for increased productivity, it remains to be seen if parallel agent usage will genuinely enhance software reliability or simply create the illusion of efficiency.