JOURNAL / RAG DOCUMENT INGESTION

How to turn PDFs into Markdown for RAG

A practical path from document upload to clean, useful retrieval context.

STARTOCR TEAM

•

6 MIN READ

KEY TAKEAWAY

Good retrieval starts with readable structure, not a larger model.

Start with a document your model can read.

PDF text is often out of order, missing headings or split across columns. Convert the document into structured Markdown before you chunk it.

Use a small, repeatable pipeline.

01
Upload the PDF
02
Request Markdown
03
Chunk by headings
04
Store embeddings

Keep the API request simple.

Use the request ID in your logs when a document needs review. This makes the pipeline easier to operate as volume grows.

POST /v1/markdown
-F file=@document.pdf
→ markdown + pages + request id

KEEP READING

Related guides

OCR, Markdown or JSON?
Choose the output that fits your application.
A simple async document workflow
Use jobs, webhooks and request logs.
How invoice data is returned
The fields your AP workflow needs.

PUT THE GUIDE TO WORK

Choose the product and workflow that matches this guide, then inspect a representative document before integrating.

Related API
Explore the document output that matches the guide.
Try an example
Open the evaluation preview before using a live request.