Felix Reiseberg, a hiring manager at Anthropic, has stopped using traditional coding interviews for job candidates, he wrote this month. He argued that Claude is now better than most people at the kinds of small coding challenges those interviews rely on, such as a short React exercise, and that the company's most impactful builders succeed regardless of how they perform on such tests. His post also sets out a path for junior engineers. He urges them to focus on business impact rather than clearing backlog tickets, citing Patrick McKenzie's 2011 line that engineers are hired to create business value, not to program things. He recommends identifying a few bold ideas that could significantly help the organisation reach its goals, then building a working prototype rather than spending weeks on requirements documents and securing buy-in. He warns against cramming several ideas into one prototype and suggests that a fresh perspective, especially on user-facing work, is an advantage. The account comes from one hiring manager's experience and is not a formal Anthropic hiring policy. It appeared in a newsletter roundup alongside other AI developments. For Australian businesses and professionals, the change itself alters little right away. Dropping a coding test costs nothing, but a practical exercise such as a short take-home task or a pair-programming session still shows how someone works. If your recruitment process uses AI tools to screen or rank applicants, those tools handle personal information, which falls under the Privacy Act 1988 and the Australian Privacy Principles. Amendments requiring transparency about automated decision-making are due to start in December 2026. It is worth trying to set junior staff small prototype projects tied to a measurable business problem, and to assess candidates on judgement, on how they review AI-generated code, and on how they explain trade-offs.