Summarize Documents: Claude + Make

A neighborhood oral-history project collects interview transcripts from volunteers, who drop them into a shared Google Drive folder. The transcripts pile up faster than anyone reads them, so the richest stories sit unopened, and the project loses track of what it already has.

This build keeps a running index. The moment a document lands in the folder, Make reads it, Claude writes a short summary, and a row is logged with the filename, the summary, and the date. The folder still holds the full documents; the sheet becomes the map of what is in them.

We wire it in Make, part of our data and research set. One limit worth flagging up front: this build reads Google Docs and text cleanly. Scanned PDFs and images need an extra text-extraction step, which the pitfalls section covers.

What does this workflow actually do?

In one line: a new document becomes a logged summary. Make watches a Drive folder. When a document appears, it reads the text, hands it to Claude, Claude writes a short summary, and Make logs the filename, summary, and date in a sheet. The originals stay in Drive; the sheet is your index.

A few real cases, none of them the usual ones:

• The oral-history project above, finally able to see what is in its archive at a glance.
• A condo board summarizing contractor proposals dropped in a shared folder before a vote.
• A graduate student turning a folder of paper drafts into a tidy literature log.
• A small law office summarizing intake memos into a running case sheet.

Collecting documents is automatic once people share a folder. Reading each one and remembering it is the grind. That reading-and-condensing is exactly what Claude is good at.

Why use Make if Claude can already read and write?

Worth asking. Claude is good at reading a document and writing a faithful short summary. It is not built to watch a Drive folder, open each file, and log rows. That is plumbing, and Make does plumbing.

Make handles the no-judgment steps: noticing the new file, fetching its text, writing the row, running on its own. Claude handles the one step that needs reading, the summary. Make is the librarian who shelves and indexes; Claude is the reader who writes the card. I tried having the AI also fetch and file, and it stumbled on where things went; letting Make move the files and Claude read them keeps the index trustworthy.

What do you need before you start?

• A Make account. The free 1,000 operations a month covers a busy folder.
• A Claude API key from the console, not the chat app. See how to use Claude.
• A Google account with a Drive folder to watch and a Sheet to log into.
• Documents in a readable format. Google Docs and text work out of the box; scanned PDFs need an extra step.

Two Make words. A scenario is the whole folder-to-log automation. A module is one box. Four boxes, because reading the document is its own step.

How does the workflow work, step by step?

Four modules, left to right:

Module	App	What it does
1. Trigger	Google Drive	Fires when a document is added to the folder
2. Read	Google Docs	Pulls the text out of the document
3. Brain	Claude	Writes a short summary
4. Output	Google Sheets	Logs filename, summary, and date

Step 1: Watch a Drive folder

Create a scenario and add Google Drive, Watch Files in a Folder. Connect your account and pick the folder people drop documents into. This trigger fires on each new file and hands the next step its name and ID.

Step 2: Read the document

Add Google Docs, Get Content of a Document and map the file ID from Step 1. This pulls the actual text so Claude has something to summarize. For Google Docs and text this works straight away; scanned PDFs need an extraction step first (see pitfalls).

Step 3: Let Claude summarize it

Add a Claude module, Create a Prompt, paste your key, pick Sonnet, and map the text from Step 2:

Summarize this document for a quick index.
Document: {{content from Step 2}}

Give five short bullet points, then one line starting "Takeaway:".
Plain language. Do not pad. If the document is very short, summarize
in fewer bullets rather than inventing detail.

Asking for bullets plus a one-line takeaway keeps the log scannable and stops Claude padding a thin document.

Step 4: Log the summary

Add Google Sheets, Add a Row, and map the filename from Step 1, Claude’s summary from Step 3, and the current date. Now the folder has a living table of contents that updates itself.

How do you run it and check the result?

Click Run once after dropping a test document in the folder. When I tested this with a few Google Docs, summaries logged within a minute, and the tuning that mattered was asking for bullets plus a one-line takeaway so the sheet stayed easy to skim. Adjust the Step 3 prompt for length, then turn the scenario on so the folder indexes itself.

After that, a folder that used to be a black hole becomes a readable index. You stop opening files to remember what they are and start scanning a summary, which for an archive that grows every week is the difference between using it and losing it.

What does this cost to run?

Piece	Free tier	If you outgrow it
Make	1,000 operations/month free	Core plan from about $9/month
Claude API	Pay per use	A document summary on Sonnet costs well under a cent
Google Drive	Free	Free with your account
Google Sheets	Free	Free

Each document uses four Make operations and a sliver of Claude. A folder that gains a few files a day sits well inside the free plan. More on tiers in our Make guide.

What can go wrong, and how do you avoid it?

• It chokes on scanned PDFs or images. The read step handles Docs and text; scans need an OCR or extraction step before Claude. Add one if your folder is full of PDFs.
• A very long document hits limits. Summarize in sections, or cap how much text you send.
• It fires on old files the first run. Normal. Clear the test rows and let it settle on new files only.
• It reads documents it should not. Point it at a folder meant for this, and keep private material out of it.

The same read-then-condense pattern runs many jobs. See Make AI scenarios for more.

How do you build this in Zapier or n8n instead?

Same four jobs, different names.

Job	Make	Zapier	n8n
Catch a new file	Watch Files in a Folder	New File in Folder trigger	Google Drive Trigger node
Read the document	Get Content of a Document	Get Document action	Google Docs node
Summarize it	Claude module	Claude (Anthropic) action	Anthropic node
Log the summary	Add a Row	Create Spreadsheet Row	Google Sheets node

Make and Zapier are easiest to start. n8n suits self-hosting. Full comparison: Zapier vs Make vs n8n.

Common questions about this build

Does it work with PDFs?

Cleanly with Google Docs and text. Scanned PDFs and images need a text-extraction or OCR step before Claude, because there is no text to read until then.

Where do the summaries go?

Into a Google Sheet as rows: filename, summary, date. You can point Step 4 at a Doc instead if you prefer a running document.

Will it summarize huge files?

Up to a point. Very long documents may exceed limits, so summarize in sections or cap the text you send in Step 2.

Can it summarize Word files from Drive?

With a conversion step, yes. The simplest path is Google Docs; for other formats, convert or extract the text first.

Is the chat app the same as the API?

Same models, different door. Make needs an API key from the console, not the chat login.

Sources and official docs

• Google Drive on Make (Help)
• Claude API overview (Anthropic)
• Google Docs basics (Google)
• Claude on Make (integrations)
• Automatic summarization — Grokipedia

Last reviewed: May 2026. Make, Claude, and Google update their interfaces; check the official pages for exact button names.

You may also like

• AI Automation for Beginners (the hub)
• Draft WordPress posts from a spreadsheet with Claude + Make
• Summarize any web page into a newsletter blurb
• Make.com Complete Beginner’s Guide
• How to use Claude AI
• Make AI Scenarios: automations that save hours
• The AI Tools Directory