IT Recruitment Must Adapt to the Era of Generative AI Development
As agentic AI shifts software engineering away from manual coding, traditional technical interviews need a complete overhaul to evaluate real problem-solving and tool usage.
Software development has fundamentally changed, yet standard hiring practices still rely on outdated testing methods. For roughly a decade, technical interviews focused heavily on unaided, manual code production under the assumption that writing syntax was the primary job of a software engineer. However, with the rise of agentic development, manual coding is increasingly becoming an exceptional state rather than the default workflow. At companies like 37signals, leadership notes that manual code writing has largely ceased to be the normal way software is built. Consequently, recruitment processes must evolve to reflect the actual environment professionals work in every day.
Rethinking Technical Assessments for Agentic Workflows
Instead of stripping away modern tools and forcing candidates into unaided, whiteboard-style coding, hiring processes should mirror modern workflows. A proper evaluation begins with a practical problem the team actually faces, complete with representative data and a reproducible environment. Candidates should be allowed to use agent harnesses, documentation, and standard development tools while being assessed on how they set goals, supply context, recognize bad directions, and decide when an output is ready for production.
Research into developer productivity and agent performance highlights the importance of measuring the complete task rather than isolated code output. Studies show that using AI assistants can alter completion times and introduce unique challenges, such as concurrency bugs or hidden logic flaws that a quick prompt might conceal. Therefore, a successful evaluation must look beyond the initial output. Reviewers should introduce failing cases or requirement changes, observing how candidates investigate defects, adapt their solutions, and verify that the final product satisfies real-world demands.
What it means for developers
For job seekers, preparing for an AI-driven interview landscape means shifting focus away from memorizing syntax or algorithms. Developers should practice taking a task from incomplete requirements to a defensible, verified result using modern tools. Developers can try top AI models cheaply through one API at https://apixoai.online. Candidates must be ready to explain their decision-making trail, including why they accepted a specific output, how they diagnosed errors, and how they managed total time and cost to reach a working solution. Employers and applicants alike benefit when interviews evaluate comprehensive technical judgment rather than simple manual typing.
Moving Toward Concrete Evidence of Judgment
Assessing modern engineers requires shifting from subjective impressions to concrete evidence. Reviewers should track specific dimensions such as problem-solving clarity, effective use of agent harnesses, technical judgment regarding risks, result quality, total time, and overall cost including retries and token expenditure. While these methods still require longitudinal validation to measure long-term employee success, moving away from archaic syntax gates is a necessary step. By aligning recruitment with how software is actually built today, organizations can better identify engineers who truly understand how to direct, verify, and deliver reliable systems.
Source: IT Recruitment in the Age of Generative AI — Towards AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
