In 2025, the AI boom sparked by ChatGPT has entered its third year. Like every technological upheaval in human history, LLMs have permanently changed how we work. Late in 2024, I watched Series A startup probook.ai search for a founding engineer by making “design an AI agent and see it through” the take-home assignment. That moment showed how leaders are tired of traditional hiring and reminded interviewers like us to reset our expectations. It’s time to rethink how we find the right talent.
Startup
You’re a startup founder. The company is growing fast, so you need experienced teammates who already work with AI. Don’t hesitate to spend on AI productivity tools. Evaluate candidates across (1) track record, (2) how they use AI productivity tools, (3) personal projects, and (4) ownership and agency.
In 2025, with coding agents already formidable, finishing a personal project with their help should be table stakes. Sourcing can stay with headhunters or job boards, but once a candidate enters the funnel, the work must speak. Open-source contributions and side projects spotlight ownership; bug troubleshooting notes or the completeness of a take-home assignment reveal AI tooling fluency. The strongest stack—Codex CLI, Claude Code, Gemini CLI, Cursor CLI—turns senior engineers into tech leads who manage multiple agents and shrinks onboarding to learning the tool stack and team rituals.
Big Tech
Plenty of enterprises already use Codex CLI, Claude Code, and Cursor CLI in daily work. When will we retire the LeetCode-style rounds? Try tools like Claude Code during interviews (for instance on ailiocode.ai) to watch how future teammates actually operate instead of rewarding short-term memory. I haven’t seen Big Tech run large-scale pilots yet, which isn’t surprising: their processes rarely change. When they finally pivot, the new system will stick around for years.
Tech Companies
I encourage all companies to keep hiring in the US, especially hiring new grads. Today’s new grads will become tomorrow’s experienced engineers who know how to employ AI agents. If you want to build a generational company and play the long game, investing early in an AI-era talent pipeline is critical.
Future Technical Interviews
The rules have shifted. Software hiring went from brainteasers to LeetCode. In the AI era, we need another way to spot talent. New failure modes show up when candidates accept every answer from an agent without judgment; interviewers must watch how they challenge and calibrate their collaborators. And yes, many companies complain that candidates use AI tools to cheat—platforms like cluely.ai make it obvious—but interviews should identify people who leverage tools best. Cheating is fundamentally an integrity check. If someone is willing to hide the assist, can they really carry ownership?