Documents & Data at Scale

Most people's relationship with AI and documents looks like this: they have a long PDF, they upload it to ChatGPT or Claude, they ask "summarize this," they get a summary. Useful? Sure. Transformational? Not really. You just saved yourself 20 minutes of reading.

Today we're going way beyond that. We're talking about processing dozens of documents at once, cross-referencing sources, building knowledge bases, and turning messy real-world data into clean, structured output.

This is Cowork's killer use case. And it might be the single most time-saving thing you learn in this entire course.

Batch processing: folders, not files

Instead of uploading one document at a time, point Cowork at an entire folder. Let it process everything.

Real examples:

"Go through every PDF in my Contract Archive folder. For each contract, extract the vendor name, contract value, start date, end date, and auto-renewal clause. Compile everything into an Excel spreadsheet with one row per contract, sorted by end date. Flag any contracts expiring in the next 90 days."

"Read every meeting notes file in my Q1 Meetings folder. Identify the top 5 recurring themes across all meetings. List every action item with the person responsible. Create a summary report with the themes at the top and the action items organized by person."

"Process every resume in my Candidates folder. For each candidate, extract their name, years of experience, key skills, and most recent company. Create a comparison spreadsheet and highlight anyone with more than 5 years of experience in product management."

One prompt. Dozens of documents. Structured output. What would take you an entire afternoon takes Cowork a few minutes.

Cross-referencing multiple sources

This is where AI starts doing something genuinely hard for humans. When you have information scattered across multiple documents, AI can hold all of it in context simultaneously and find connections you'd miss.

"Read the three analyst reports in my Research folder and my internal sales data spreadsheet. Where do the analysts' market predictions align with our actual sales trends? Where do they contradict? Create a report highlighting the agreements and disagreements, with specific data points from each source."

You'd spend hours flipping between documents trying to hold all the threads in your head. AI reads everything at once and maps the connections instantly.

Building knowledge bases

Here's a concept that's incredibly powerful and doesn't require any technical knowledge to use.

Imagine you could take everything your company has ever written — memos, reports, meeting notes, strategy docs, emails — and have an AI that "knows" all of it and can answer questions about any of it.

That's essentially what you're doing when you point Cowork at a folder of documents and start asking questions. The AI reads everything and can draw on all of it.

The way this works behind the scenes is simple: the AI retrieves relevant pieces of your documents and uses them to ground its answers in your actual data, rather than making things up. You don't need to understand the technical details. What matters is the practical outcome — you can ask questions like "what did we decide about pricing in Q3?" and get an answer drawn from your actual meeting notes and strategy docs, with the AI pointing you to the specific source.

The more organized your files are (remember Day 5, when we set up Cowork with clean folder access?), the better this works. Good file hygiene isn't just neat. It's the foundation of a useful knowledge base.

AI with spreadsheets — real analysis

When most people use AI with spreadsheets, they ask for a chart or a summary. That's barely scratching the surface.

Cowork can do real analytical work with your data:

• Calculate trends, averages, outliers, and growth rates
• Build formulas and pivot-table-style analyses
• Cross-reference data across multiple sheets or files
• Create new spreadsheets with processed and categorized data
• Flag anomalies and patterns you didn't ask about

The key is being specific about what you're looking for. Instead of "analyze this spreadsheet," try "analyze month-over-month revenue by product line, calculate growth rates, identify which products are declining, and flag any month where a product dropped more than 15% from the prior month."

Specific questions get specific answers. Vague questions get summaries you could've written yourself.

Turning chaos into structure

This is the pattern underlying everything in today's lesson. You have messy, unstructured, real-world information — scattered documents, inconsistent formats, data in different places — and you need it clean, organized, and actionable.

That transformation, from chaos to structure, is the task humans are worst at and AI is best at. We get bored. We lose focus. We miss things in document 47 that contradicted something in document 12. AI doesn't have those problems.

The next time you're staring at a pile of information thinking "ugh, I need to go through all of this," stop. That's not your job anymore. That's Cowork's job. Your job is to review the structured output and make decisions based on it.

Today's exercise: process a real folder (15 minutes)

Find a folder on your computer with at least 5-10 documents in it. Could be anything: receipts, meeting notes, reports, client files, research papers, contracts.

Open Cowork. Point it at the folder. Ask it to extract specific information from every document and compile it into a structured spreadsheet or summary report.

Be specific about:

• What data points to extract from each document
• What format the output should be in (spreadsheet, report, list)
• Where to save the result
• Any sorting, filtering, or flagging you want

Review the output. Check a few entries against the original documents to make sure they're accurate. This is your quality control step — trust but verify, especially the first time you run a new workflow.

If the results are good, you just built a reusable workflow. Next time you have a similar pile of documents, the same prompt works again. Or schedule it (Day 8) to run automatically on a folder you add to regularly.

What's coming tomorrow

Tomorrow we're connecting AI to your existing tools — not just Cowork, but browser extensions, automation platforms, and integrations that put AI everywhere you already work. The goal: zero friction between "I need help" and "AI is helping."

AI doesn't just help you work with documents. It eliminates the worst part of knowledge work — the tedious extraction of meaning from chaos.