流程驱动，还是人才密度？

Lately, working on projects with some big companies, I kept feeling progress was slow and stuck somewhere, without knowing why, until I saw Li Xiang's take. He divides companies into two models. The first is process-driven. These firms handle extremely complex categories, like the barrel theory: one weak stave and the whole business fails. Phones, operating systems, electric cars cannot afford a single dropped detail, so companies like Huawei, Xiaomi and IBM run very strict processes to keep every stave level. But such companies often feel heavy when doing internet or AI, which need fast trial and error and flexible organisation. The second is talent-density driven, like ByteDance and Google. They rely not on process but on a high concentration of talent, tied together by a simple tool like OKRs and pulled forward by top people's own drive. This suits platform-level innovation and frontier research, especially AI and large models, because such organisations can find a path through chaos. They tend to be weaker at hardware, which needs countless details along the chain to fit together. So asking a traditional process company to do internet or AI is a mismatch. In the AI era, things change so fast that today's logic may be overturned tomorrow. I increasingly agree with keeping the organisation small while keeping talent density high: short decision chains, and every person counts. Instead of doing everything yourself, partner to bring in outside resources and stay light. Uncertainty may be opportunity, not just risk.