Anthropic Manager Drops Coding Interviews as AI Outperforms Candidates in Programming Tests

Anthropic, the artificial intelligence company behind Claude, is facing a major shift in how software engineers are evaluated after one of its engineering managers decided to remove traditional coding interviews from his team’s hiring process. Felix Rieseberg, who has worked on products including Claude.ai, Claude Code Desktop and Cowork, announced the decision in a blog post published on October 27, 2026. He argued that conventional programming exercises have become less effective because AI tools can now solve many of the small coding problems candidates are asked to complete during interviews. Rieseberg explained that these tests previously offered a useful, although imperfect, indication of a candidate’s ability to solve technical problems. However, he believes their value has declined as AI systems become increasingly capable of writing code, answering programming questions and completing common software engineering tasks. His decision highlights a broader debate within the technology industry about whether hiring practices developed before modern AI tools can still accurately measure a developer’s potential. Importantly, this is a decision by one manager for his team, not an announcement that Anthropic has abolished coding interviews across the entire company.

Rieseberg’s argument reflects the rapid improvement of AI coding assistants, which can generate applications, identify bugs, explain unfamiliar code and suggest solutions to complex technical problems. Tools such as Claude Code are increasingly capable of handling tasks that once required developers to spend considerable time writing and debugging code manually. As a result, a candidate’s performance on a short programming challenge may no longer provide a complete picture of how effectively that person would work in a real engineering environment. Rieseberg argues that some of the most productive developers on his teams are not necessarily the people who achieve the highest scores on traditional coding tests. Instead, their value comes from recognising important problems, making sound decisions and delivering useful solutions. This does not mean that programming knowledge has become irrelevant. Developers still need to understand software architecture, test results, evaluate AI-generated code and identify security or performance problems. Rather, the emphasis is shifting towards how effectively engineers combine technical understanding with AI tools to produce reliable outcomes.

One particularly interesting aspect of Rieseberg’s announcement is his optimistic view of opportunities for junior developers. Although AI automation has raised concerns that entry-level software jobs could become harder to obtain, he believes the technology can also help ambitious newcomers compete with more experienced engineers. His advice is to focus on solving meaningful business problems rather than simply completing routine assignments or accumulating years of experience with particular programming frameworks. He recommends identifying a problem that matters to an organisation, developing one ambitious idea and building a working prototype over a weekend. A prototype allows a developer to demonstrate what an idea can accomplish rather than relying entirely on presentations or written proposals. Rieseberg cited an example involving two relatively new Anthropic engineers, Anthony Morris and Alice Zhao, who reportedly developed a prototype graphical interface for Claude Code over a weekend. According to his account, the first public version was released approximately eight days later. The example illustrates how AI-assisted development can reduce the time required to turn an idea into a working product, potentially allowing developers with less experience to demonstrate their abilities through practical results.

However, the shift also raises difficult questions about the future of recruitment and career development. If employers rely less on conventional coding tests, they will need alternative ways to evaluate candidates’ technical competence, problem-solving ability and responsible use of AI. Rieseberg acknowledged that he has not yet established a complete replacement for the traditional coding interview. For now, he says he is paying closer attention to résumés, portfolios and small prototypes that demonstrate good judgment, empathy for users and thoughtful design decisions. His approach also comes amid wider changes in the technology industry. Google’s engineering teams have reportedly been testing interviews that allow candidates to use Gemini, while other technology companies are reconsidering how they measure productivity in AI-assisted workplaces. Anthropic’s own October 2026 employment research also reported that junior-level workers had become a smaller share of its technical workforce, with junior employees accounting for 14% of technical staff, down from 21% in early 2025. The share of junior hires had fallen from 27% to 9%. Although these figures do not establish that AI alone caused the changes, they underline the importance of developing new pathways for people entering the technology profession.

For the wider software industry, the announcement signals a possible transformation in what employers value when recruiting engineers. As AI systems take over more routine programming tasks, companies may increasingly seek people who can identify worthwhile opportunities, direct AI agents, evaluate their output and take responsibility for the final product. This could create opportunities for developers who learn to use AI effectively, even if they have not spent many years working with particular technologies. At the same time, it could make the traditional route into software engineering more challenging if junior employees lose opportunities to learn through basic assignments and entry-level positions become less common. Aspiring developers should therefore continue building strong programming foundations while learning how to use AI coding tools, test applications, identify errors and develop projects that solve genuine problems. Rieseberg’s decision does not prove that coding interviews are obsolete everywhere, but it demonstrates that established hiring practices are being reconsidered as AI capabilities improve. The key question for employers will be how to distinguish candidates who can merely generate code with AI from those who can use it to build secure, reliable and valuable software.