Split PDFs with Natural Language Instructions

Tell the AI how to split in plain English. It reads document content, detects boundaries, and names each file - things page-based splitters could never do.

• AI-powered boundary detection - with custom instructions
• 3 split modes - smart, fixed intervals, bookmarks
• Batch processing - multiple PDFs in one request
• Auto-naming - split files named by content

Just need to split a few files? Use the free web tool

Why AI splitting beats page-based tools

Page-based splitters count pages. AI reads content and understands boundaries.

Quickstart

Two endpoints: upload a file directly, or pass a URL and we'll fetch it. Use mode=smart with instructions to tell the AI how to detect boundaries.

curl -X POST https://www.renamed.to/api/v1/pdf-split \  -H "Authorization: Bearer YOUR_API_TOKEN" \  -F "file=@mixed_documents.pdf" \  -F "mode=smart" \  -F 'instructions=Split when you detect a new document.Look for: invoice headers, "Page 1 of X", or new letterhead.Keep cover letters with their attachments.'

curl -X POST https://www.renamed.to/api/v1/pdf-split/from-url \  -H "Authorization: Bearer YOUR_API_TOKEN" \  -H "Content-Type: application/json" \  -d '{    "url": "https://example.com/scanned-stack.pdf",    "mode": "smart",    "instructions": "Separate by patient name.\nEach patient'"'"'s records should be one file.\nInclude all pages until you see a new patient header."  }'

SDKs

Use our official SDKs for a better developer experience with type safety and error handling.

import { RenamedClient } from '@renamed/sdk'const client = new RenamedClient({ apiKey: process.env.RENAMED_API_KEY })// AI reads the bundle, detects boundaries, names each splitconst job = await client.pdfSplit('./mixed_invoices.pdf', {  mode: 'smart',  instructions: `Split when you detect a new invoice.Look for: company letterhead, "Invoice #", or "Page 1 of X".Each vendor's invoice should be separate.`})// Poll for completionconst result = await job.waitForCompletion()console.log(result.splits)// => [{ filename: "2025-01-15_AcmeCorp_Invoice_001.pdf", reason: "New invoice detected - Acme Corp header" }, ...]

{  "jobId": "abc123",  "status": "completed",  "progress": 100,  "results": {    "results": [{      "originalFilename": "mixed_documents.pdf",      "splits": [{        "filename": "2025-01-15_AcmeCorp_Invoice_001.pdf",        "downloadUrl": "https://...",        "expiresAt": "2025-01-16T12:00:00Z",        "pages": 3,        "reason": "New invoice detected - Acme Corp header"      },      {        "filename": "2025-01-14_GlobeTech_Invoice_002.pdf",        "downloadUrl": "https://...",        "expiresAt": "2025-01-16T12:00:00Z",        "pages": 2,        "reason": "New vendor letterhead - GlobeTech Inc"      }]    }]  },  "creditsUsed": 2}

SDKs on GitHub: github.com/renamed-to/renamed-sdk

Authentication

All API requests require authentication via Bearer token in the Authorization header.

API Parameters

Key parameters for controlling the split behavior.

What your instructions can do

The AI interprets natural language and applies complex boundary detection that would require custom code to build yourself.

Boundary detection

"Split when you see 'Page 1 of X' or a new company letterhead"

Document grouping

"Keep cover letters with their attachments. Don't split appendices from main documents."

Entity-based splitting

"Separate by patient name - each patient's records should be one file"

Header detection

"Each PURCHASE ORDER starts a new document. Invoice headers also indicate new documents."

Split modes

Choose the mode that fits your use case. Smart mode is the default.

Built for production

Getting started

Four steps to integrate the PDF Split API into your workflow.

FAQ

How does AI boundary detection work?

In "smart" mode, the AI analyzes document content to identify where one document ends and another begins. It detects changes in letterhead, document type, vendor, "Page 1 of X" patterns, and structural elements. You can guide it with natural language instructions.

What can I put in the instructions parameter?

Natural language describing how to detect boundaries. Examples: "Split when you see a new invoice header or Page 1 of X", "Each document starts with a cover letter followed by attachments", "Separate by patient name - each patient's records should be one file".

How long does processing take?

Most jobs complete in 30 seconds to 2 minutes depending on file size and page count. The API is async - you submit a job and poll for completion.

What's the max file size?

PDF files up to 100MB. There is no hard page limit, but very large documents may take longer to process.

How are split files named?

In smart mode, the AI generates descriptive filenames based on document content - extracting dates, vendors, document types, and other entities. Each split also includes a "reason" field explaining why the boundary was detected.

How is this different from traditional PDF splitters?

Traditional splitters work on page numbers or bookmarks. This API reads document content, understands meaning, and applies your natural language instructions. It can split by vendor, patient, document type, or any semantic boundary - things impossible with page-based tools.