Any Sphere 联合创始人兼 CEO Michael Tru 探讨了 Cursor 的使命——创造一种超越代码的新型编程方式，并分享了关于快速增长和反直觉经验的心得。
Michael Tru, co-founder and CEO of Any Sphere, discusses Cursor's mission to create a new type of programming that moves beyond code, and shares insights on rapid growth and counterintuitive lessons.
要点 · TL;DR
编程将从写代码演进为表述意图，人类始终掌握主导权。 Programming will evolve from code to intent specification, keeping humans in control.
Cursor 的核心魅力来自定制模型，而这是他们最初未曾设想的。 Cursor's key magic comes from custom models, a direction they didn't initially expect.
AI 对软件开发的影响将超过互联网，并持续数十年。 AI will transform software development more than the internet, over a span of decades.
核心观点 · Key points
编程的未来将远离代码，转向更高层次的意图描述，人类仍将处于主导地位。 The future of programming will move away from code to higher-level intent specification, with humans in the driver's seat.
在软件工程中，品味和逻辑设计技能将比谨慎更加重要。 Taste and logic design skills will become more valuable than carefulness in software engineering.
定制模型对产品质量至关重要，尤其是在速度和成本方面。 Custom models are essential for product quality, especially for speed and cost reasons.
成功使用AI工具需要将任务分解成小块并频繁迭代。 Successful AI tool usage involves chopping tasks into smaller pieces and iterating frequently.
招聘应专注于世界级人才，即使耗时更长，并采用工作测试进行评估。 Hiring should focus on world-class talent, even if it takes longer, and use work tests for evaluation.
AI的转变将比互联网更具影响力，并将持续数十年。 The AI shift will be more consequential than the internet and will take decades to unfold.
反共识 · Contrarian takes
我们最初没预料到会自己开发模型，但现在Cursor的每个神奇时刻都涉及定制模型。 We didn't expect to do our own model development, but now every magic moment in Cursor involves a custom model.
我们招聘太慢，而不是太快，这与对初创企业的常见建议相反。 We hired too slow, not too fast, contrary to common advice for startups.
增长初期感觉缓慢，因为指数增长起点低，即使是历史性产品也是如此。 The growth felt slow initially because exponential growth starts low, even for a historic product.
IDE市场比过去的编辑器市场大得多，因为AI能创造更多价值。 The IDE market is much bigger than past editor markets because AI enables far more value creation.
像微软这样的现有企业可能在这个市场中挣扎，因为市场对它们不友好。 Incumbents like Microsoft may struggle in this market because it's not friendly to them.
资深工程师往往低估AI的潜力，而初级工程师则过度依赖AI。 Senior engineers often underrate AI's potential, while junior engineers over-rely on it.
本期章节 · Chapters（共 30）
引言Introduction
赞助插播Sponsor Break
欢迎与愿景Welcome and Vision
软件开发未来Future of Software Development
Cursor的起源Origin of Cursor
IDE与模型与智能体IDE vs Model vs Agentic Product
AI工程师与IDE未来Future of AI Engineers and IDE
AI时代的工程管理Engineering Management with AI
初次打造CursorBuilding Cursor for the First Time
从零构建CursorBuilding Cursor from scratch
One Schema赞助One Schema sponsor segment
反直觉：自建模型Introduction and Sponsor
自建模型的意外Counterintuitive Learning: Building Own Models
模型集成与开源Surprise About Own Models and Stack
AI的可防御性Model Ensemble and Open Source
市场规模与赢家Defensibility in AI
微软与CopilotMarket Size and Winner
Cursor用户建议Microsoft and Copilot
重建直觉Tips for Cursor Users
Cursor对初高级工程师影响Rebuilding Intuition
给过去自己的建议Impact of Cursor on Junior vs Senior Engineers
招聘教训Advice to Past Self at Cursor
两天面试项目Hiring Lessons Learned
AI热潮中保持专注Two-Day Interview Project
对AI进展的误解Staying Focused Amid AI Hype
AI变革关键玩家Misconceptions about AI progress
Cursor招聘需求Key players in the AI shift
工程角色未来Hiring needs at Cursor
软件未来与工程需求Future of engineering roles
Future of Software and Engineering DemandFuture of Software and Engineering Demand