Learn Documents to Spreadsheet | Understanding Cowork

Swipe to show menu

Core Concept

This is the flagship extraction workflow - a pile of receipts, invoices, or contracts turned into one clean spreadsheet in minutes. The critical difference from organize: you define the schema. Cowork cannot guess your columns. You spell them out, and Cowork fills them in.

How It Works (Step-by-Step)

Step 1 - Mount the folder

A folder of documents. Receipts, invoices, contracts - anything with structured information buried in unstructured layout. Cowork course students can use the sample receipts/ folder as a starting point.

Step 2 - DEFINE the schema out loud

This is the step everyone skips and everyone regrets. State exactly what columns you want, and any edge-case rules.

"I want all of these in one spreadsheet. Columns: date, vendor, category, amount, currency, payment method, source filename. Mark uncertain cells with a question mark. If a receipt is in a different currency, keep the original amount and add a converted-to-AUD column."

Notice the last two - those are edge case rules Cowork wouldn't have known without you telling it.

Step 3 - Watch READ

Cowork processes each document. On a folder of eighteen receipts, this takes under a minute. Comment out loud on what it's doing - this is the moment where the plan becomes actual output.

Step 4 - Read the proposed spreadsheet

Cowork will surface the extracted data as a preview. Read the schema - every column you specified should be there. Read the rows - spot-check a few. And look for the source column. Every row should have a filename in it so you can trace any value back to its origin.

Step 5 - VERIFY on a real row

The trust step. Pick one row - a specific amount, a specific date, a specific vendor. Open the source file. Confirm the value. This is what makes extraction a trustworthy workflow instead of a hopeful one.

Step 6 - Catch an edge case

There will be one. A missing field, a currency it wasn't sure about, a date format it flagged. Show the edge case, tell Cowork how to resolve it, watch it update.

Step 7 - EXPORT

Cowork saves the spreadsheet as .xlsx (or .csv if you prefer). Open the file. Scroll it. Done.

Why This Matters

Extraction is where most knowledge workers see the biggest single time-saving from Cowork. Anywhere you have documents with structured data buried in unstructured layout - receipts, invoices, contracts, resumes, applications, form submissions - this pattern turns hours of manual data entry into minutes of prompted extraction.

What Should Be Possible After This

Define an extraction schema up front, including edge cases.

Read Cowork's proposed spreadsheet critically and spot rows to verify.

Verify a row against its source document.

Handle currency mixes, format ambiguities, and missing fields.

Export a clean .xlsx ready to send.

Study More

Common Pitfalls

Not defining the schema. Vague requests produce vague spreadsheets. Be specific about columns.
Skipping the source column. Without a source-per-row, you can't audit the extraction. Always ask for it.
Trusting the output without verification. Cowork is very good at extraction. Very good is not perfect. Spot-check anywhere numbers or names carry stakes.

1. Which step in the Claude Cowork extraction workflow is essential for making sure the spreadsheet columns match exactly what you want?

2. Drag and drop each key step of the Claude Cowork extraction workflow into the correct order below.

Everything was clear?

Thanks for your feedback!

Section 1. Chapter 11

Ask AI

Ask anything or try one of the suggested questions to begin our chat

Section 1. Chapter 11