用 ChatGPT Work 插件核对财报,自动发现数据不一致
Use ChatGPT Work to deliver board-ready reporting
So, as you can see, I'm in the chat GPT um window right now. I'm going to navigate to the work tab here. Um, and I'm going to get ready to get started with my audit work. So, I've got my quarter end materials, a revenue memo for my executives, a board package, as well as my financial uh model. What I'm doing right now is I'm using the plugins functionality to bring all that data right into chatbt work. So I prompt I say we just closed the quarter please review the latest financial materials housed in my Google drive and so it's pulling all that context into chatgbt work and take a check against what's been posted in Netswuite in our underlying financial system.
所以,如你所见,我现在在ChatGPT的窗口里。我要导航到工作标签页,准备开始我的审计工作。我这里有季度末材料、给高管的收入备忘录、董事会资料包,还有我的财务模型。我现在正在做的是利用插件功能,把所有这些数据直接导入ChatGPT工作区。所以我提示说,我们刚结束了季度,请审查我Google Drive中的最新财务材料,然后它就把所有上下文拉入ChatGPT工作区,并与我们在底层财务系统Netsuite中发布的内容进行核对。
There's a lot of numbers thousands of rules uh you know multiple tabs and it's come up with two key critical issues and giving me unfortunately a no-go decision. So to my team let's figure out how to fix this. Um, the first issue it spotted is that Netswuite actually posted an update to the actuals in June. We're also showing an outdated forecast that's flowing through my my revenue memo what versus what's housed in my model.
有大量的数字、成千上万的规则,还有多个标签页,它发现了两个关键问题,并给了我一个不幸的“不可行”决定。所以,我的团队,让我们想办法解决这个问题。第一个问题是,Netsuite实际上在6月发布了实际数据的更新。我们还显示了一个过时的预测,这个预测贯穿了我的收入备忘录,与我的模型中的内容不一致。
What I like about this workflow is that chat GBT helps guide me through how to fix these errors, edit it one by one. I can use the browser functionality embedded right within chat to open this up and send a note or comment to my team to ask them please fix. Um, and I can tab around to each of the tabs that has have been impacted by this error. [gasps] A lot of these documents, of course, are pretty dense. Let's look at my uh memo that I prepared for my executives.
我喜欢这个工作流程的地方在于,ChatGPT帮助我指导如何修复这些错误,逐一编辑。我可以使用Chat内嵌的浏览器功能打开这个,并给我的团队发送一条便条或评论,请他们修复。而且我可以在受此错误影响的各个标签页之间切换。这些文档当然很多都很密集。让我们看看我为高管准备的备忘录。
This is 20 page long. Total a lot, lots of text, very easy to miss a number in all this dense uh documentation, but chatbt actually goes ahead and sort of highlights the areas in red, so I know exactly where to update uh I sort of press my updates. The cool thing about this is once I've gone through and sort of asked the team to send the updates, I can rerun this again and again and and in minutes I'll get a quick answer in terms of if we're ready to send.
这份备忘录有20页长。内容很多,文字密集,很容易在这些密集的文档中漏掉一个数字,但ChatGPT实际上会继续用红色标出这些区域,所以我知道确切需要更新的地方。我按下更新。酷的是,一旦我完成并让团队发送更新,我可以一遍又一遍地重新运行,几分钟内就能得到快速答案,告诉我们是否准备好发送。
But I think what it really gives me is the confidence to sort. And I think the the thing about using these plugins is this example can actually be extended beyond this. So think about, you know, my board deck which is you know 34 pages getting slotted into a broader deck board deck with hundreds of pages. I can use the same skill to cross-ch checkck numbers, but also surface themes that might be sort of conflicting across the different documentations.
但我认为它真正给我的是排序的信心。我认为使用这些插件的好处是,这个例子实际上可以扩展到其他方面。所以想想我的董事会演示文稿,它有34页,被插入到一个更广泛的、有数百页的董事会演示文稿中。我可以使用同样的技能来交叉核对数字,同时也能发现不同文档中可能相互冲突的主题。
So, it's a great skill to get started and a really good one to just make sure we're confident in the numbers we're [clears throat] putting out there to the world.
所以,这是一个很好的入门技能,也是一个确保我们对发布给世界的数字充满信心的好方法。
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力