AI‑native interview round
- Sierra cofounder Bret Taylor said the company replaced traditional coding rounds with a two‑hour 'plan‑build‑review' session that allows AI tool use. - The session evaluates product thinking, technical judgment, and sensible AI usage during a hands-on build phase. - This interview change signals some firms are prioritising AI fluency, product judgment, and real‑world building over isolated algorithm puzzles (Bret Taylor on X).
Sierra has replaced traditional coding interviews with a two-hour “plan-build-review” session that lets candidates use artificial intelligence tools. (x.com) Bret Taylor, Sierra’s cofounder and chief executive, said the round tests how candidates scope a problem, build something working, and explain their choices in a review. Taylor said the company is looking for product thinking, technical judgment, and “sensible” use of AI during the exercise. (x.com) Taylor is not a fringe voice in hiring. He is Sierra’s chief executive, OpenAI’s board chair, and a former co-chief executive of Salesforce; Sierra was valued at $10 billion in a September 2025 funding round led by Greenoaks. (cnbc.com) Sierra sells AI agents for customer service, and Taylor has been arguing publicly that software is shifting from clicking through interfaces to describing tasks in plain language. At HumanX in April 2026, he said Sierra had already launched Ghostwriter, a tool meant to build other agents from prompts. (techcrunch.com) That makes the interview format part of the same bet. A company building AI agents is screening engineers in a setting where using AI is normal, not treated as cheating or a shortcut. (x.com) The old model Sierra is moving away from is the timed coding round built around algorithm puzzles, the kind of prep market served by sites like LeetCode. Those tests can measure fluency with data structures and edge cases, but they do not closely resemble a day spent shipping product with modern coding tools. (leetcode.com) Other companies are also formalizing how candidates can use AI in hiring, though the rules differ. Anthropic says candidates may use Claude for some parts of the process and only when explicitly allowed in assessments, while OpenAI’s interview guide emphasizes collaboration, communication, and role-specific evaluation rather than a single standard coding format. (anthropic.com) (openai.com) Sierra is also still hiring early-career engineers through its APX rotational program, which TechCrunch reported in September 2025 was aimed at recent technical graduates and mixed agent engineering with product management. Taylor said those roles were designed to give new hires “an irresponsible amount of responsibility.” (techcrunch.com) The practical question for candidates is no longer just whether they can solve a puzzle on a whiteboard. At Sierra, the test now starts with a plan, moves into a build, and ends with a review of how they used the tools in front of them. (x.com)
