Simple A.I. to Work with Excel Spreadsheet --- Breakthrough
- Jack Lau
- 1天前
- 讀畢需時 4 分鐘
Last night, almost out of curiosity, I installed Claude for Excel.
I wasn’t trying to run a demo or prove a point. I simply had a real task at hand:to compare the financial performance of Hong Kong’s UGC-funded universities—revenue, expenditure, surplus, and the trend across multiple academic years.
Normally, this would mean:
jumping between annual reports,
copying figures from PDFs and news articles,
cleaning inconsistent tables,
and manually building comparisons in Excel.
Instead, I tried something different.
Excel + AI, But Done the Right Way
What Claude for Excel enables is not “AI inside Excel” in a cosmetic sense.It creates a working loop between three things:
External data crawlingClaude can search for publicly available information (e.g. audited reports, UGC summaries, media coverage).
Data organization and extractionIt understands what the spreadsheet is asking for:
revenue vs expenditure
year-on-year surplus
institution-level comparison
trend analysis
Context-aware interaction with the sheet itselfThe AI does not work in isolation.It reads the structure of the spreadsheet, fills in missing pieces, flags unavailable data, and keeps consistency across rows and years.
This is the critical difference.
A Concrete Example: UGC-Funded Universities
In my case, I asked Claude to help me build a comparison across UGC-funded institutions:
HKU
CUHK
HKUST
PolyU
CityU
HKBU
Lingnan
EdUHK
Within a very short time, I had:
a structured table of income, expenditure, and surplus,
multi-year surplus trends,
notes on data availability (where figures were not yet public),
and a chart that actually reflected the financial trajectory rather than just one-year snapshots.
This is not trivial automation.It is analysis-aware spreadsheet construction.
Why This Matters More Than It Looks
find the data,
decide what matters,
and tell it exactly what to compute.
Claude for Excel changes that dynamic.
Now, Excel becomes:
a thinking workspace,
where AI can search, interpret, reconcile, and explain,
while still preserving the rigor and transparency of a spreadsheet.
You can see the sources.You can see the numbers.You can challenge assumptions.
This is especially important for:
finance and policy analysis
education funding
institutional benchmarking
board-level preparation
A Subtle but Important Shift
This is not about replacing analysts.It is about compressing the distance between question and insight.
When AI can:
crawl external data,
organize it coherently,
and reason inside Excel,
the spreadsheet stops being a static reporting tool and becomes a live analytical environment.
That, to me, is the real breakthrough.
If you want to install: https://claude.com/claude-in-excel
Claude for Excel：當試算表開始「思考」
昨晚，我幾乎是出於好奇，安裝了 Claude for Excel。
並不是為了做示範，也不是想證明什麼。我只是剛好有一個實際的工作需要完成：比較香港教資會（UGC）資助大學的財務表現——收入、支出、盈餘，以及跨年度的變化趨勢。
以往，這類工作通常意味著：
在多份年報與新聞之間來回切換，
從 PDF 手動抄數，
清理格式不一致的表格，
再慢慢在 Excel 裡拼湊比較。
這一次，我換了一種方式。
Excel 與 AI 的正確結合方式
Claude for Excel 並不是「把 AI 塞進 Excel」那麼簡單。它真正建立的是一個 三者互動的工作流程：
對外部資料的搜尋與爬梳Claude 能主動查找公開資料，例如：審計報告、UGC 彙總、媒體分析等。
資料理解與結構化它理解你在問什麼：
收入 vs 支出
年度盈餘
不同院校的橫向比較
長期趨勢，而不只是單一年份
與試算表本身的語境互動AI 不是在旁邊「建議」。它會讀懂試算表的結構、補齊缺漏、標示尚未公開的數據，並維持跨年度的一致性。
這一點，非常關鍵。
一個具體例子：UGC 資助大學
在我的案例中，我要求 Claude 協助比較以下院校：
港大（HKU）
中大（CUHK）
科大（HKUST）
理大（PolyU）
城大（CityU）
浸大（HKBU）
嶺大（Lingnan）
教大（EdUHK）
很短時間內，我已經得到：
結構清晰的收入、支出與盈餘表格，
多年度盈餘趨勢分析，
清楚標註哪些數據尚未公開，
以及真正反映財務走勢的圖表，而非單點數字。
這已經不只是自動化，而是具備分析意識的試算表建構。
為什麼這件事很重要
多年來，Excel 一直很強大，但它是「被動的」。它等待人類去：
找資料，
判斷重點，
明確告訴它要算什麼。
Claude for Excel 改變了這個關係。
Excel 開始成為：
一個可思考的工作空間，
AI 能搜尋、理解、整理、解釋，
同時保留試算表應有的嚴謹與透明。
你仍然看得到數據來源，仍然能質疑假設，仍然能自己下判斷。
這對以下場景特別重要：
財務與政策分析
教育資源配置
機構比較
董事會與管理層準備
一個不張揚，但深遠的轉變
這不是要取代分析師。而是大幅縮短「問題」與「洞察」之間的距離。
當 AI 能夠：
主動取得外部資料，
有條理地整理，
並且在 Excel 內部推理，
試算表就不再只是靜態報表工具，而是一個即時的分析環境。
對我來說，這才是真正值得注意的突破。
如果你想安裝，可以到這裡：
(*i am beginning to use A.I. to help me with the writing. Hopefully, i can bring more new new things to all of us.)
留言