Dewu and Greentown Detail Enterprise AI Verification Practices
In an interview published by TMTPost, Dewu and Greentown China described contrasting paths for deploying enterprise AI, with the former starting from financial analysis and the latter planning at group level. Their practices included code-based calculations, multi-agent cross-checking, risk-tiered human review, document-level source tracing and continuous evaluation. Both companies emphasized data governance, business participation and ongoing post-launch testing.
The report offers concrete implementation practices for making enterprise AI outputs more verifiable and maintainable, including multi-agent review, source attribution, risk-based controls and continuous testing.
在绿城中国,业务人员向AI提出的问题正在发生变化。最初,他们想知道上个季度销量的数字,现在,他们希望AI找出表现不佳的项目并分析原因,再从企业积累的案例中寻找改善办法。
随着问题的而深入,业务人员对AI的要求也越来越高。所以,数据口径、知识更新、结果核验和部门协作,都成了AI能否继续用下去的具体条件。
11月11日,由TMT研究院与ACCA联合发起的《从数字化到AI》系列内容第三期上线。
本期嘉宾是得物高级财务总监瞿孜文与绿城中国数字化建设中心总经理补声东,一位从财务场景推进AI应用,一位从集团层面统筹数字化建设,两条路径的交汇处,让我们看到了企业如何让AI的输出经得起长期价值检验的关键。
不同行业的AI推进策略不同
绿城先做集团层面的AI规划,再由各部门结合自身业务形成具体方案。随着一线尝试增多,数字化团队得以了解业务真正想解决的问题,也找到能够共同推进应用的人。