Automate PDF to Markdown with n8n, Zapier, and Make

Converting a single PDF to Markdown is a two-second job. Converting the hundredth PDF this month, by hand, one file at a time, is a chore you should never have to do. The moment PDFs arrive on a schedule, from an inbox, a shared drive, or a form upload, the smart move is to stop touching them at all. Let an automation platform watch for new files, convert each one, and drop clean Markdown wherever it needs to go.

This guide shows you how to build that pipeline in the three most popular no-code and low-code tools: n8n, Zapier, and Make. The shape of the workflow is the same in all three. A trigger notices a new PDF, an HTTP step sends it to a conversion API, and one or more output steps store the resulting Markdown in your notes app, your repository, or a vector store for AI retrieval. Once it is running, PDFs turn into Markdown on autopilot while you do something more useful.

The Anatomy of a PDF-to-Markdown Automation

Every automation you build here, no matter the platform, has the same four parts. Understand these once and you can rebuild the flow anywhere.

1. Trigger: Something announces a new PDF. A file appears in a Dropbox or Google Drive folder, an email arrives with an attachment, or a webhook fires from your own app.
2. Fetch the file: Download the PDF binary so the next step can send it onward. Most platforms hand you the file content directly from the trigger.
3. Convert: Send the PDF to a conversion endpoint and receive Markdown back. This is the one step that does the real work.
4. Route the output: Take the Markdown and put it somewhere useful: a Notion page, a Git commit, a database row, or a vector store for retrieval-augmented generation.

The conversion step is the heart of it. You need an endpoint that accepts a PDF and returns clean Markdown with headings, lists, and tables intact. PDFtoMD exposes exactly that kind of API, and any tool that can make an HTTP request can call it. If you would rather see the request and response shape in detail before wiring it into a platform, the PDF to Markdown API integration guide walks through the raw calls step by step.

Why Automate Instead of Converting by Hand

Manual conversion is fine for the occasional file. It stops being fine the moment any of these are true:

• Volume: You receive more than a handful of PDFs a week. Invoices, reports, research papers, contracts, and statements pile up fast.
• Consistency: Every file needs the same treatment: same output folder, same naming, same downstream step. Humans forget. Workflows do not.
• Speed to value: A PDF that lands at 2 a.m. should be searchable Markdown by 2:01 a.m., not whenever someone gets to it.
• Feeding AI systems: If you are populating a knowledge base or a vector store, the ingestion has to be hands-off to stay current.

Automation turns a recurring task into infrastructure. You build it once, and it quietly works forever.

Building the Pipeline in n8n

n8n is the most flexible of the three because it is open source, self-hostable, and gives you a real HTTP Request node with full control over headers and body. That makes it ideal for calling a conversion API directly.

Step 1: Choose a trigger

Add a trigger node that fires on a new file. Common choices are the Google Drive Trigger (watch a folder), the Dropbox node polling for new files, or an IMAP Email node that reads attachments. Configure it to watch the exact folder or mailbox where your PDFs land.

Step 2: Download the PDF binary

Most triggers give you a file ID rather than the bytes. Add the matching Download node (for example, Google Drive to Download) so the next node has the actual PDF as binary data attached to the item.

Step 3: Send it to the conversion API

Add an HTTP Request node. Set the method to POST, point it at your conversion endpoint, add your API key as a header, and send the PDF as multipart form data. A typical configuration looks like this:

Method: POST URL: https://pdftomd.cloud/api/convert Authentication: Header Auth Name: Authorization Value: Bearer YOUR_API_KEY Body Content Type: Form-Data (multipart) Field name: file Type: n8n Binary File Input data field: data

The response comes back as JSON with the Markdown inside it. In later nodes you reference it with an expression such as the JSON path to the markdown field on the HTTP node output. n8n makes that easy through its expression editor, so you never have to hand-parse anything.

Step 4: Route the Markdown

Now decide where it goes. Add one or more of these output nodes:

• Write to Google Drive or Dropbox: Save a .md file next to the original.
• Create a Notion page: Push the Markdown into a database as a new entry.
• Insert into a database: Store the text in Postgres, Airtable, or Supabase.
• Upsert into a vector store: Chunk the Markdown and send it to Pinecone, Qdrant, or Weaviate for AI retrieval.

Because n8n runs the whole graph per file, a folder of fifty PDFs becomes fifty clean Markdown documents with no clicks from you.

Building the Pipeline in Zapier

Zapier is the friendliest option if you do not want to host anything. Its strength is the enormous library of app connectors. Its constraint is that steps run one item at a time, so it suits steady trickles of files rather than giant batch dumps.

Step 1: Pick the trigger

Create a new Zap and choose a trigger like New File in Folder (Google Drive or Dropbox) or New Attachment in Gmail. Zapier will hand the following steps a file object it can pass along.

Step 2: Convert with a Webhooks action

Add a Webhooks by Zapier action, set it to POST, and configure it to send the file to your conversion API:

• URL: your conversion endpoint, for example the PDFtoMD convert route.
• Payload Type: Form.
• File: map the file field from the trigger step into the request.
• Headers: add your Authorization header with the API key.

Zapier automatically parses the JSON response, so the Markdown becomes a field you can reference in later steps by name.

Step 3: Deliver the result

Add a final action to store the Markdown. Popular endpoints include Create Page in Notion, Create File in Google Drive, Add Row in Airtable, or Send Channel Message in Slack so a human gets notified with the converted text attached. If you are collecting research notes, sending the output straight into Notion or Obsidian is a natural fit. See the Notion import guide for how the Markdown maps to blocks once it arrives.

Building the Pipeline in Make

Make (formerly Integromat) sits between the other two. It has a visual canvas like Zapier but supports richer data handling, routers, and iterators, which makes it strong for fan-out flows where one PDF should land in several places at once.

Step 1: Add a watch module

Start a new scenario with a Watch Files module for Google Drive or Dropbox, or Watch Emails for a mailbox. Set the folder and the polling interval.

Step 2: Download and convert

Add a Download a File module to fetch the binary, then an HTTP: Make a request module to send it onward:

Method: POST URL: https://pdftomd.cloud/api/convert Headers: Authorization: Bearer YOUR_API_KEY Body type: multipart/form-data Fields: file: [the file from the Download module]

Turn on Parse response so Make reads the JSON automatically and exposes the Markdown as a usable field in the next modules.

Step 3: Fan out with a router

This is where Make shines. Add a Router after the HTTP module and branch the Markdown into several destinations at once: one route creates a Notion page, another commits the file to a Git repository through the GitHub module, and a third pushes chunks into a vector store. One incoming PDF, three synchronized outputs, zero manual steps.

Comparing the Three Platforms

There is no wrong choice. Pick the one you already use, or the one whose pricing fits your volume. The conversion step is identical in each.

Feeding a Vector Store on Autopilot

The most valuable version of this pipeline ends in a vector store. Once your PDFs become clean Markdown, you can chunk them, embed the chunks, and make years of documents searchable by an AI assistant. Markdown matters here because its headings and lists give your chunker natural boundaries, which produces cleaner segments and better retrieval than raw extracted text. If you are building this kind of system, the RAG pipelines guide explains why the format has such a large effect on retrieval accuracy.

The automation is straightforward. After the conversion step, add a code or function step that splits the Markdown on its headings, then an embedding call, then an upsert into your vector database. From then on, every PDF that arrives is automatically indexed and ready for question answering.

Practical Tips for Reliable Automations

• Name outputs predictably: Derive the Markdown filename from the original PDF name so you can always trace a document back to its source.
• Handle errors: Add an error branch that alerts you (an email or a Slack message) when a conversion fails, so a bad file never disappears silently.
• Deduplicate: Track processed file IDs in a database or the platform data store so the same PDF is never converted twice.
• Mind the polling interval: Tighter intervals mean faster conversions but more operations consumed on your plan. Match it to how urgently you need the output.
• Keep the original: Store the source PDF alongside the Markdown for audit or re-conversion later.

These are the differences between a demo that works once and a pipeline you can trust for months without looking at it. For a wider view of where automated conversion fits across teams and industries, browse the use cases to see how others structure their document flows.

Frequently Asked Questions

Do I need to write any code to build this?

No. In Zapier and Make, every step is a configured module, including the HTTP call to the conversion API. n8n is nearly as visual, though it rewards you for being comfortable with expressions. The only place code helps is the optional chunking step before a vector store, and even that is a short snippet.

Which platform is best for large batches?

n8n handles high volume most gracefully because it processes items in a graph and can run self-hosted without per-task pricing. Make is a strong second for batches. Zapier processes one item per run, so very large dumps can get expensive, though it is perfect for steady, moderate flows.

Can I convert PDFs that arrive as email attachments?

Yes. All three platforms have email triggers that expose attachments. Point the trigger at a dedicated inbox, filter for PDF attachments, and feed each one into the conversion step. This is a common pattern for invoices and statements.

What happens to tables and headings?

A good conversion preserves heading hierarchy, lists, and tables as proper Markdown, which is exactly what makes the output useful downstream. That structure is why Markdown beats plain text extraction for both human reading and AI ingestion.

How much does the conversion step cost to start?

You can begin on the PDFtoMD free tier, which includes three conversions a month with no credit card required. That is enough to build and test your entire pipeline end to end before you commit to a paid plan for higher volume.

The Takeaway

Automating PDF to Markdown is one of those small pieces of infrastructure that quietly pays off every day. Whether you build it in n8n, Zapier, or Make, the recipe is the same: watch for a new file, send it to a conversion API, and route the clean Markdown to wherever it belongs. Set it up once and the manual conversion chore disappears for good.

Start with the platform you already know, wire in the conversion step, and point it at a single folder. The first time a PDF turns itself into searchable Markdown while you were not even watching, you will wonder why you ever did it by hand.