多数企业的AI预算都流向了成果好量化的销售和市场部门，少有重投在像财务、采购等后 端部门，原因很简单，这些部门虽然能真正因AI数字化带来降本，可成果却很难被量化。 在生成式AI技术爆发后，各行业对AI有一个共识，它是一个需要算力喂养和技术迭代的辅助决策工 具，它能降本增效。 "钱小事大。"韩格盈介绍，云海肴覆盖北京、上海等30多个城市的150余家直营门店，每个门店发生的 每一笔单子金额都不大，但整体数量大，导致财务审批工作极为繁琐，她希望AI能尽快介入团队的工 作流程中。 企业落地应用哪套AI系统背后，有一场成本暗战：现在不投怕落后，投了又怕见不到效果。 北京合思信息技术有限公司的创始人兼CEO马春荃，在SaaS领域创业11年，其创立的敏捷财务收支管理 平台合思，如今为云海肴、蓝月亮、泡泡玛特等消费领域企业提供"财务场景+AI"的解决方案，还是科 大讯飞、北森等科技企业背后的AI费控服务商。 在过去四五年里，马春荃发现，多数企业的AI预算都流向了成果好量化的销售和市场部门，少有重投 在像财务、采购等后端部门，原因很简单，这些部门虽然能真正因AI数字化带来降本，可成果却很难 被量化。 审批报销流程更快，合规风险 ...

Core Insights - The majority of companies are directing their AI budgets towards sales and marketing departments, where results are easier to quantify, while less investment is seen in back-end departments like finance and procurement, despite the potential for cost reduction through AI [1][4][5] - There is a consensus across industries that generative AI serves as a decision-making tool that requires computational power and technological iteration, ultimately leading to cost reduction and efficiency improvement [2] Group 1: AI Investment Trends - Companies are hesitant to invest in AI for finance and procurement due to the difficulty in quantifying the results, even though these areas can benefit from cost reductions [1][4] - The founder of a SaaS company noted that many enterprises are caught in a cost dilemma: they fear falling behind if they do not invest, yet worry about not seeing tangible results from their investments [3][4] Group 2: AI Implementation Challenges - The finance department is often the most cautious in adopting AI due to high compliance, accuracy, and data security requirements, leading to a slower pace of AI integration compared to marketing and sales [5][6] - Many companies are still uncertain about which data can be accessed by AI, contributing to a lag in AI transformation within finance [5][6] Group 3: Case Study - Cloudy Yao - Cloudy Yao, a restaurant chain, has a significant portion of its finance team dedicated to expense approvals, with each employee reviewing over 500 invoices monthly, highlighting the cumbersome nature of the process [6] - After implementing AI for expense approvals, Cloudy Yao reported an average time savings of 4345 minutes per invoice, equating to three days, and an 80% reduction in approval error rates [6][7] Group 4: Future of AI in Business - The market expert emphasized that while CFOs are eager to embrace new technologies, quantifying the financial impact of AI investments remains challenging due to the lack of standardized pricing and the shift towards results-based payment models [7][8] - A report from MIT indicated that 95% of global AI investments have not generated economic benefits, trapping companies in a cycle of high investment with zero returns [8]