DeepSeek founder Liang Wenfeng says his workers don't do overtime or even have KPIs

Liang Wenfeng, the founder of the artificial intelligence firm DeepSeek, has publicly addressed the internal operational structure of his company. Furthermore, he noted that the organization has moved away from traditional corporate metrics, explicitly stating that his workers do not have key performance indicators, commonly referred to as KPIs, to measure their output or success. The coverage from Fortune emphasizes the contrast between DeepSeek's internal labor practices and the broader expectations of the technology industry. By highlighting the absence of both mandatory extra hours and rigid performance metrics, the reporting underscores a specific managerial philosophy adopted by Liang Wenfeng.

The focus remains on the founder's own assertions regarding how the workforce is managed and the lack of standard corporate pressure tactics typically found in high-growth AI development environments. This development arrives at a time when the global technology sector is often characterized by intense competition and demanding work schedules. The mention of a lack of KPIs is particularly significant because these metrics are standard tools used by most major corporations to track efficiency and goal attainment. Liang Wenfeng's decision to omit these structures suggests a departure from conventional business administration, focusing instead on a model that avoids the traditional pressures of performance tracking and extended working hours.

Future attention will likely center on how this lack of formal performance metrics impacts the long-term productivity and output of DeepSeek. Based on the facts provided by Fortune, the company is currently operating under this non-KPI, no-overtime framework. Observers will be watching to see if this management style is maintained as the company scales or if the absence of these traditional metrics leads to changes in how the founder oversees his workforce and the development of their AI technologies.