最近,Microsoft 365上线了Claude in Excel,继Claude Code让程序员们充满压力后,Claude in Excel也注定会给分析师带来很多影响。相比国内厂商们毫无新意地卷着AI生成PPT、AI写邮件、AI生成简历...不得不说微软和Claude这波操作真是既实用又让人眼前一亮。让我们看一下Claude in Excel的各种实用功能:1、数据清洗:面对脏数据,你只需要一句clean up the data,无论是异常值、错误值、不规范的值统统可以修正。2、改正公式:同样只需要一句debug the formulas in the sheet。3、搭建简单模型:I have a $600k 20 year mortgage with5% interest. Create an amortizationtable breaking down the outstandingbalance and the interest paymentbreakdown (principal vs interest).4、搭建复杂模型:把Apple的年报上传,让他根据年报做财务预测,并且给出假设。using apples most recent annual report i've attached, build a clean3 statement model and forecast it forward 5 years. use reasonable assumptions where needed, keep inputs separate and make sure there are no errors。然后模型就做好了
,抛开财务模型不谈,Claude in Excel几乎准确的从苹果的财报里提取出了利润表、资产负债表、现金流量表,在以前对于小编而言,这是非常繁琐且易错的工作,不停地copy、paste和抄数据...。当然Claude 给出来的假设看上去还挺像那么回事的。Excel对于分析师好比VS Code这种集成开发环境对于程序员,虽然都是工具,但是承载着大家工作的思路和方法,不过这下看上去,无论是脏活累活、还是所谓的脑力劳动,Claude都可以在一定程度上帮助你了,不知道这是喜是悲
看到Claude in Excel时,我第一反应是看看有没有豆包 in 飞书多维表格,毕竟小编工作中也经常用到飞书多维表格、Coze这些产品,也用的非常顺手,但是小编并没有找到这么一个类似的插件。不过我还是比较了一下国内模型和Claude的效果,把同样的问题给了豆包和通义千问。豆包分业务板块给出了营收预测,并且给出了盈利性预测等Revenue Growth
- iPhone: 3-5% annual growth (driven by Pro model demand and emerging markets; FY2025 growth = 4%)
- Mac: 4-6% annual growth (stable enterprise and consumer demand; FY2025 growth = 12%)
- iPad: 3-5% annual growth (education and hybrid work tailwinds; FY2025 growth = 5%)
- Wearables, Home and Accessories: 0-2% annual growth (recovery from FY2025 decline; FY2025 growth = -4%)
- Services: 10-12% annual growth (advertising, App Store, and cloud services expansion; FY2025 growth = 14%)
Profitability
- Products Gross Margin: 36.5-37.0% (stable cost structure offsetting tariff risks; FY2025 = 36.8%)
- Services Gross Margin: 75.0-76.0% (scalable model with low incremental costs; FY2025 = 75.4%)
- R&D Expense % of Revenue: 8% (consistent investment; FY2025 = 8%)
- SG&A Expense % of Revenue: 7% (efficient operations; FY2025 = 7%)
- Effective Tax Rate: 15-16% (foreign earnings mix and tax credits; FY2025 = 15.6%)
当然,豆包同样对于财报的三张表的数据做了提取和相应的预测2、再看通义:通义同样也从年报中提取出了数据,但是通义的预测和估算明显更粗线条一点Key Design Principles
Clean structure: All inputs are separated from calculations.
Integrated statements: Income Statement → Cash Flow Statement → Balance Sheet.
Reasonable assumptions: Based on historical trends, management commentary, and industry context.
No circular references or hard-coded numbers in formulas
看来通义还挺实在
我想做一款类似于lovart的产品,功能类似,现在需要搭建一个财务模型:
一、核心假设(Inputs - 可调整)(用户与定价策略(参考 Lovart)、用户增长假设(5年预测)、AI Token 成本假设(关键变量)、其他成本假设)
二、收入测算模型(Annual)
三、成本测算模型(Annual)
四、利润与关键指标(Year 3)
五、模型扩展建议
AI 首先冲击的,是程序员、分析师、技术文档工程师这些高学历、高薪资的“知识精英”。在AI工具越来越丰富的今天,每个人都需要思考如何利用AI保值,如何进行再技能化。AI 的浪潮不会退去,但它冲刷出的不是平坦的沙滩,而是新的高地与深谷。
决定你职业命运的,也不是 AI 有多强,而是你选择让它扮演什么角色。
内容参考:https://www.youtube.com/watch?v=ydCeHVYSvCI行业 | 前沿技术 | 应用
携手共建AI技术交流社区
