让出门问问盈利的一场内部革命：裁员、降薪、取消中层

Core Viewpoint - The company is undergoing a significant internal transformation, termed the "reduction revolution," aimed at achieving profitability through cost-cutting measures, including layoffs and salary reductions, while also focusing on AI-driven management tools [2][3][4]. Group 1: Financial Performance - The company reported a significant increase in revenue, with per capita revenue rising from 542,000 yuan in 2024 to 978,000 yuan in 2025, nearly doubling [12]. - The company's R&D expenditure was reduced from 55 million yuan to 38 million yuan, with salaries decreasing from 46 million yuan to 21 million yuan [3]. Group 2: Organizational Changes - The "reduction revolution" involves streamlining business lines and reducing headcount from nearly 400 to under 200, with a two-thirds reduction in R&D personnel [3][4]. - The founder, Li Zhifei, is implementing a management philosophy that emphasizes transparency and direct oversight of engineering work through a custom-designed software called Coding Agent [5][9]. Group 3: AI Integration - The Coding Agent software serves as the operational core of the company, aiming to transform management and organizational structures through AI [4][10]. - The company is pursuing an "AI organizational transformation," which includes eliminating middle management layers to enhance direct communication between executives and frontline employees [10][11]. Group 4: Management Philosophy - Li Zhifei acknowledges his management shortcomings and is using his technical background to create a management tool that aligns with his vision of an efficient, AI-driven organization [4][7]. - The management approach has shifted from subjective performance evaluations to a system where every action taken by employees can be tracked and assessed through the Coding Agent [9][10]. Group 5: Challenges and Future Outlook - Despite positive early results, the company has not fully eliminated middle management, indicating that the transformation is still ongoing [14]. - The Coding Agent has improved efficiency but also introduced new challenges, such as the time-consuming nature of fixing code issues [15].