Once spending tens of millions daily on "shrimp farming," Wang Puchong discusses the four phases of Meituan's AI application.

Recently, Wang Puzhong, CEO of Meituan's core local business, shared his views on the application of AI in enterprises during a public speech, along with insights into Meituan's internal AI transformation process. Wang stated that currently, while the supply of AI is abundant, only a few companies have been able to measure tangible benefits from its implementation. This issue stems from a fourfold mismatch: in terms of understanding, companies either exaggerate or underestimate AI; in terms of efficiency, large models are applied to routine tasks without touching the core; in terms of scenarios, AI still hovers on the periphery, failing to integrate into the core operations; and in terms of assessment, there is a pursuit of immediate results while neglecting the internal adaptability of the enterprise. According to Wang, the accumulation of these four mismatches has made it difficult for AI investments to measurable productivity growth. Wang revealed that Meituan's AI transformation has gone through four stages. The first stage, from February to March of this year, was marked by the company initiating a "shrimp farming campaign" for all employees, but the result was skyrocketing bills, with daily expenses exceeding ten million yuan, while the errors generated by shrimp farming interfered with genuine operations.