Automate Microsoft Excel with Lapu AI

Top Microsoft Excel workflows

1. 1. Clean and reconcile a messy spreadsheet end-to-end

   You receive an export from another team with mixed date formats, trailing whitespace, inconsistent capitalization, and duplicate rows. Lapu AI opens the file, profiles every column, proposes a cleanup plan, and applies it after you approve — leaving an audit row that lists every transformation.

   "Open ~/Downloads/q2-pipeline.xlsx, profile each column for type and quality issues (dates as strings, trailing spaces, duplicates by 'deal_id'), show me the cleanup plan, then apply the fixes and save as ~/Downloads/q2-pipeline-clean.xlsx. Add a 'Changelog' sheet listing every transformation."

   Tools used: file:read, file:write, doc:excel, shell:exec

2. 2. Generate a monthly report from local CSV exports

   Every month you assemble a report from three CSV exports — one from Stripe, one from HubSpot, one from your CRM. Lapu AI reads each, joins on customer ID, builds the summary sheet with the formulas and pivot tables you specified, and saves the workbook in your reports folder.

   "Read every CSV in ~/exports/2026-05/, join rows by 'customer_id', and build a workbook at ~/reports/2026-05-revenue.xlsx with three sheets: 'Raw', 'Monthly Summary' (SUMIFS by plan), and 'Top 50 Accounts' (sorted by ARR). Format the summary sheet with bold headers and currency formatting on the $ columns."

   Tools used: file:read, doc:excel, shell:exec

3. 3. Write and explain complex formulas in plain English

   You know what you want — 'sum sales where the rep is in the West region and the deal closed in Q2' — but XLOOKUP, SUMIFS, and array formula syntax is fiddly. Lapu AI inspects your sheet structure, writes the formula, drops it into the right cell, and explains what each part does so you can adapt it later.

   "Open ~/deals.xlsx, look at the 'Pipeline' sheet structure, then in cell H2 of the 'Summary' sheet write a SUMIFS that totals 'Amount' where Region = 'West' AND Close Date is in Q2 2026. After inserting it, paste a plain-English explanation as a comment on the cell."

   Tools used: doc:excel, file:read

4. 4. Bulk-rename and reorganize a folder of workbooks

   Your shared drive has 200 client workbooks named 'untitled-1.xlsx', 'final-v2-USE-THIS.xlsx', etc. Lapu AI inspects each file's content (sheet names, top rows, client field), generates a clean name like '2026-Q1_AcmeCorp_Invoice.xlsx', and moves files into subfolders by client.

   "Walk ~/Dropbox/clients/, open each .xlsx, read cell B1 of the first sheet for the client name and cell B2 for the period, then rename to '{period}_{client}_{originalCategory}.xlsx' and move into ~/Dropbox/clients/{client}/. Show me the proposed rename map before doing anything."

   Tools used: file:read, file:write, shell:exec, doc:excel

5. 5. Convert PDF tables into a clean Excel workbook

   A vendor sends a 40-page PDF statement and you need it as a working spreadsheet. Lapu AI extracts every table from the PDF, normalizes columns across pages, validates totals against the printed page footer, and writes the result to a new .xlsx.

   "Read ~/Downloads/vendor-statement-may.pdf, extract every line-item table, combine into one sheet, and save as ~/finance/vendor-statement-may.xlsx. Validate that the sum of the 'Amount' column matches the 'Total' on the last PDF page; flag any mismatch in cell A1."

   Tools used: file:read, doc:pdf, doc:excel, file:write

6. 6. Update a budget tracker from a bank CSV

   You categorize transactions in a personal or team budget workbook. Lapu AI reads the latest bank export, matches each row against your category rules sheet, appends new rows to the tracker, and highlights anything it could not classify.

   "Read ~/Downloads/transactions-may.csv, look at the 'Rules' sheet in ~/finance/budget.xlsx for category mappings (vendor → category), append matched rows to the 'Transactions' sheet with the date, vendor, amount, and category, and highlight unmatched rows in yellow so I can categorize them manually."

   Tools used: file:read, doc:excel, file:write

7. 7. Refresh a chart-heavy dashboard with new data

   You have a dashboard workbook with charts and pivot tables built on a 'Data' sheet. Lapu AI replaces the 'Data' sheet with this week's export, refreshes formulas and pivots, and saves a snapshot copy with the date in the filename.

   "Open ~/dashboards/sales-dashboard.xlsx, replace the contents of the 'Data' sheet with the contents of ~/exports/sales-2026-05-26.csv, refresh all pivot tables and formulas, and save a copy as ~/dashboards/snapshots/sales-dashboard-2026-05-26.xlsx without changing the original."

   Tools used: file:read, doc:excel, file:write, shell:exec

How it works

1. 1

   Choose the right path: file edits or driving the app

   For most bulk work — reading, writing, formulas, pivots, formatting — Lapu AI edits the .xlsx or .xlsm file directly using the open Office Open XML format that Excel has used since 2007. For tasks that require the Excel desktop UI (e.g. chart rendering quirks, Excel-specific dialogs, COPILOT function calls), the agent drives the Excel app through OS accessibility APIs the same way a human would click.

2. 2

   Read your local files

   Lapu AI reads .xlsx, .xlsm, .xls, and .csv files from your filesystem. Workbooks stay on your machine — the agent never uploads them to a Lapu AI cloud. When a task needs to reason about the data, only the relevant cells are sent as context to AI model providers, not the entire workbook.

3. 3

   Plan before editing

   For any edit beyond reading, the agent first writes a plan: which sheet, which cells, which formula or formatting change, and what the new file path will be. You approve the plan, then watch as it executes. This mirrors how Microsoft's own Agent Mode in Excel works — a step-by-step plan you can review before anything changes.

4. 4

   Preserve formulas, formatting, and macros

   When editing .xlsm macro-enabled workbooks, the agent preserves existing VBA macros rather than stripping them. When editing .xlsx files with formulas, it writes new formulas using Excel-native syntax (SUMIFS, XLOOKUP, FILTER, LET) so the file behaves identically whether opened on macOS, Windows, or Excel for the web.

5. 5

   Save with an audit trail

   Every workbook edit is recorded — which file, which sheet, what changed, what the result was. The audit trail lives on your machine and is retained for up to 90 days. For destructive edits (overwriting an existing file, deleting a sheet), the agent always saves a backup or asks for explicit approval first.

Permissions required

• file:read — to open .xlsx, .xlsm, .xls, and .csv files from your filesystem
• doc:excel — built-in spreadsheet skill: read cells and ranges, write formulas, create sheets, apply formatting, build pivot tables, refresh charts
• file:write — to save modified workbooks back to disk (the agent prefers writing to a new file unless you tell it to overwrite)
• shell:exec — optional, for batch operations like find/mv on large folders of workbooks or running Python scripts that complement Excel processing
• browser:control — optional, only needed when driving the Excel desktop app UI for tasks the file format alone cannot do (e.g. Excel-specific dialogs, exporting to PDF via Excel's renderer)

FAQ

Do I need Microsoft 365 Copilot to use Lapu AI with Excel?

No. Microsoft 365 Copilot and its Agent Mode in Excel are separate paid features from Microsoft, tied to a Copilot license. Lapu AI is independent — its desktop agent edits .xlsx and .xlsm files directly using the open Office Open XML format, and drives the Excel app through OS automation when needed. You only need a working copy of Excel (any supported version on macOS or Windows) or just the files themselves.

Will Lapu AI break my macros or formulas?

No. When editing .xlsm macro-enabled workbooks, the agent preserves existing VBA macros rather than rewriting them. When changing formulas, it uses native Excel syntax (SUMIFS, XLOOKUP, FILTER, LET) so files behave identically across Excel for Mac, Windows, and the web. For destructive edits, the agent always saves to a new file or asks before overwriting.

Can Lapu AI work with .xlsx files that are not open in Excel?

Yes — that's the common path. Lapu AI reads and writes .xlsx and .xlsm files directly via their Office Open XML structure, so most bulk work (cleanup, reconciliation, report generation) does not require Excel to be running. The agent only opens the Excel app when a task needs UI features (chart rendering, Excel-specific exports, or interacting with the COPILOT function in a live workbook).

How is this different from Microsoft's Agent Mode in Excel?

Microsoft's Agent Mode (generally available on Excel for Windows, Mac, and web for Copilot subscribers as of 2026) edits the workbook you're currently viewing in Excel. Lapu AI's desktop agent works across your whole machine — it can open multiple workbooks, read CSVs and PDFs as inputs, run shell commands, hit your CRM in a browser tab, and write the consolidated result back to a folder of Excel files. Excel's Agent Mode lives inside one workbook; Lapu lives on your desktop.

Is my spreadsheet data sent to Lapu AI's cloud?

No. The agent reads and writes Excel files directly on your filesystem. They never go to a Lapu AI server. When the agent needs to reason about a cell or formula, only the specific context the model needs is sent to AI model providers — never the full workbook as bulk storage. For the privacy-conscious, this is the main reason to choose Lapu over a SaaS spreadsheet AI.

Can Lapu AI handle massive workbooks (100K+ rows)?

Yes, with the right approach. For large files, the agent reads the workbook in chunks rather than loading every row into model context. It uses Python-style libraries that handle .xlsx files efficiently (the same Office Open XML format pandas and openpyxl work with). For workbooks above a few hundred thousand rows, expect the agent to suggest CSV intermediates or a chunked plan rather than holding the whole file in memory.

Does Lapu AI work with Excel on macOS as well as Windows?

Yes. Lapu AI ships native builds for macOS and Windows, and Microsoft Excel runs on both. Because the agent operates on the file format and via OS-level automation, it doesn't depend on platform-specific Excel features — the same workflows run on both operating systems.

When should I NOT use Lapu AI for Excel work?

Use a hosted ETL service (Fivetran, Airbyte) or a scheduled cloud function if you need server-side, 24/7 spreadsheet pipelines without a desktop session. Use Power Automate if you're deep in the Microsoft 365 stack and want flows that trigger on Outlook events or SharePoint changes. Lapu AI is built for interactive, permissioned, desktop-first Excel work — not background production data pipelines.

Related

• See all Lapu AI integrations →
• Browse use cases →
• Pricing plans →
• Lapu AI vs alternatives →