Hunger over tenure: What Anthropic’s end of the coding interview signals for IT hiring

Hunger beats tenure. That is the line at the center of a short post published this week by Felix Rieseberg, an engineering manager at Anthropic who has led engineering for Claude.ai, Cowork and Claude Code Desktop. Rieseberg wrote that he has stopped giving candidates coding interviews: a small React challenge now tells him no more than a quiz on assembly language would, because Claude is better at both. His most impactful builders, he says, succeed regardless of how they score on such tests.

One hiring manager at one AI company is an anecdote, and a conflicted one: Anthropic sells the technology in question, and readers should weigh the claim accordingly. But the practice he is retiring, and the argument he builds on top of it, partly line up with survey and labor-market data that IT hiring leaders would be unwise to ignore. The same data also shows where the argument gets complicated: hunger can only beat tenure if newcomers are hired in the first place.

What “hunger beats tenure” actually claims
The more interesting claim isn’t about interviews. It’s that impact is now “less bounded” by years of accumulated framework knowledge, and that newer engineers are overtaking more experienced and arguably more skilled coders. In his telling, comparatively new engineers are flying past people who are probably better coders and have been around much longer.

Rieseberg’s advice for engineers without status or organisational power is practical. Tie your work to business outcomes rather than the backlog. Pick one radical idea. Build a working prototype over a weekend rather than lobbying for a product requirements document. Resist stuffing twenty ideas into one demo.

His evidence is internal. He says two relatively new Anthropic engineers built a working prototype of a graphical interface for Claude Code over a single weekend, and that a public version shipped about eight days later. His underlying thesis is that execution is no longer the expensive part of software, which breaks the traditional bargain in which junior staff earn the right to have ideas by grinding through tickets first.

For a CIO, that is a structural claim worth testing against data, not a hiring fad.

Why tenure’s edge is shrinking
The adoption side of the story is well documented. In April 2024, Gartner predicted that 75% of enterprise software engineers would use AI code assistants by 2028, up from fewer than 10% in early 2023. Its survey of 598 organisations in the third quarter of 2023 found that 63% were already piloting, deploying or running such tools.

Gartner has since pushed further on what the job becomes. Philip Walsh, a senior principal analyst at the firm, has said that building AI-empowered software will demand “a new breed of software professional, the AI engineer,” and Gartner has estimated that roughly four in five engineers will need to upskill by 2027. Walsh also describes an “AI-first” mindset in which engineers mainly steer AI agents toward the right context and constraints. If the job is steering agents rather than recalling framework details, years of accumulated syntax count for less than curiosity and judgment.

The agent wave is arriving unevenly. Figures attributed to Gartner and circulated in industry coverage suggest 40% of enterprise applications will include task-specific AI agents by the end of 2026, yet only 17% of organisations had deployed agents at the time of its 2026 CIO survey, with 60% expecting to within two years.The gap between ambition and deployment is exactly where a weekend prototype can matter.

The catch: hunger needs a door to walk through
Rieseberg’s optimism lands against a more worrying backdrop for early-career workers. Stanford’s Digital Economy Lab, using ADP payroll data, found that employment for 22-to-25-year-olds in the most AI-exposed occupations fell by roughly 13% relative to others, and by nearly 20% for young software developers measured from their late-2022 peak. Workers aged 26 to 55 in the same occupations were stable or growing. Co-author Erik Brynjolfsson has said that where AI augments rather than automates work, young workers’ employment held up or grew.

The researchers are careful about causation. A February 2026 follow-up from the lab noted that under stricter controls the relative decline becomes significant only from 2024, and that interest rates don’t explain the entry-level pattern. Co-author Bharat Chandar has also said he still believes AI’s overall effect on aggregate employment is small, even as he updated his view on young workers.

Gartner’s own data points the same direction:
• A fourth-quarter 2025 Gartner survey of 110 heads of HR found that 22% reported at least one business leader who had stopped hiring for entry-level roles because of AI automation. The same survey found 95% of organisations had implemented AI in some form, but only one in five had realized significant or transformational value.
• In a separate survey of 509 supply chain leaders, 55% expected agentic AI to reduce entry-level hiring.

That last prediction is from Gartner’s supply chain practice, not its software engineering team, so treat it as an analogy rather than a forecast for IT. But the mechanism it describes, cutting the bottom rung and then paying more to buy talent later, applies cleanly to engineering.

There are also cautionary data points on whether the cuts pay off. Gartner research cited by LeadDev found that among organisations piloting autonomous business capabilities, 80% reported workforce reductions, yet they weren’t tracking a measurable ROI impact. Gartner analysts also indicated that agents tend to create more work for humans, not less. Quinn Slack, CEO of Sourcegraph, put the junior-developer dilemma bluntly: if AI can fix small bugs with little human help, why pay a junior to do it?

What IT hiring should change

Screen for hunger, not tenure. If algorithmic puzzles no longer separate strong builders from weak ones, they’re a poor filter. Rieseberg says he is now scanning resumes and portfolios for small prototypes that show taste, empathy and principled thinking, while admitting he hasn’t yet found a reliable replacement. Expect that gap to be filled by work samples, paired build sessions and review of AI-assisted output, and expect to iterate.

Don’t let the pipeline run dry. The Gartner warning and the Stanford findings together describe a risk of hollowing out your future senior ranks. Hiring fewer juniors may be rational this quarter and costly in 2030. Consider smaller cohorts hired for judgment and product sense rather than ticket throughput.

Give newcomers a way to be heard. Rieseberg’s most transferable point is organisational. Most companies are built around a few people approving or blocking ideas, and junior staff are expected to execute faster rather than aim at business goals. If prototypes are now cheap, give newcomers a sanctioned route to show one to leadership, and ask what friction they notice before they stop noticing it. Rieseberg, a self-described longtime critic of promotion by tenure, would also have you look hard at any gate that makes seniority the price of being heard.

Measure value, not just cost. Walsh has long argued that traditional ROI frameworks centered on cost reduction miss most of the value of AI code assistants. The same logic applies to talent decisions: a team that ships a validated prototype in a weekend is generating option value that headcount math won’t show.

Invest in review capacity. Gartner has noted that organisations will need to focus on reviewing the output of AI-augmented tools, and practitioners warn that AI-generated code can accelerate software debt. Senior judgment becomes more valuable, not less, which is another reason to keep apprenticeship alive.

Caveats worth stating
The evidence base is thinner than the headlines. Stanford’s authors caution that factors other than generative AI may play a role, and the aggregate labor-market effect appears small so far. Survey figures on entry-level hiring come from small samples and, in Gartner’s case, from specific practices such as HR and supply chain. A survey of 250 IT leaders by Reveal IT, reported by LeadDev, found nearly half of companies reporting AI-driven job creation versus 18% reporting AI-related layoffs, which complicates the layoff narrative.

And Rieseberg’s own claim, that hunger now beats tenure, rests on his observations at a company that builds AI tools and employs unusually strong engineers. It may not transfer to a regulated bank or a legacy-heavy manufacturer.

Hunger over tenure, with conditions
Still, the direction is consistent across sources. Adoption of AI coding tools is heading toward near-universal at enterprises. Entry-level hiring is being reconsidered in AI-exposed functions. And the skills that distinguish good engineers are shifting from recall of frameworks toward problem selection, taste and the ability to steer and verify machine output.

The coding interview may be the first ritual of the old model to go. The harder question for IT leaders is which ones come next, and whether they replace them deliberately, so hunger gets a real chance to beat tenure, or let the talent market do it for them.