Twenty files in. One clean context out.

The attachment limit problem

ChatGPT allows up to five files per session. Perplexity caps uploads by plan tier. Gemini, Claude, and Copilot all impose similar limits — and they are not the kind of thing you can negotiate. If your research requires synthesizing twelve papers, analyzing nine contracts, or comparing twenty supplier documents, you cannot upload them all.

The usual workaround is to merge the files externally, which usually means uploading them to a server you do not control. For a research paper that is fine. For a contract, a patient record, or an internal company document, it is not.

Autonom Context Merge runs entirely in your browser. Drop the files in, get a single cleaned document out, upload that one file to whatever AI tool you use. Nothing crosses a network.

What “clean” actually means here

When you extract text from a PDF, you do not get prose. You get a layout. Lines break at visual boundaries rather than paragraph endings. Words hyphenate across lines. Page numbers, running headers, and footers appear as if they were body text. Lists lose their markers. Headings get merged into the paragraph beneath them.

Autonom cleans each document before merging, so the merged output reads like a properly typeset book — one continuous document with real paragraphs, real headings, real lists:

• Hyphenated word joining. infor- then mation becomes information. Genuine hyphens like state-of-the-art are never affected because they never break across lines.
• Paragraph reflow. Broken visual lines are joined into real paragraphs. Blank lines — the true paragraph boundaries — are preserved.
• Page number removal. Standalone digits, Page 3 of 12, 3 / 12, and - 3 - are removed.
• Running header detection. Lines that repeat three or more times within a document are stripped. Detection is per-document, so a header from one file does not affect another.
• List preservation. Lines starting with -, *, •, or a numbered marker stay on their own lines.
• Heading preservation. Short lines with no trailing punctuation are treated as headings, not merged into body text.

How the merge works

Every document is cleaned independently first — so a running header in file A does not get confused with a similar line in file B. Then the cleaned documents are concatenated with a visible separator. You choose the separator style:

• Filename heading — each document starts with its filename as a heading. Best when the AI needs to reference sources by name (“In the contract dated 2024-03, clause 4.2 states…”).
• Dividing rule — a plain horizontal rule between sections. Neutral, works well for homogeneous content.
• No separator — documents are joined end-to-end with a single blank line. Best for very short files that should read as one continuous stream.

Why this saves tokens too

Cleaning a single document saves almost nothing — a few tokens here and there. Cleaning twenty documents saves a lot. Every removed page number, every rejoined line break, every stripped header is a token you do not pay for. Typical savings across a merged multi-document set are 10–25%, which translates directly into longer usable context per session.

The larger savings are structural: instead of pasting twenty attachments across multiple sessions (and losing conversational continuity), you paste one context block into one session and ask questions across the entire set.

Supported file types

• PDF — text extracted locally using PDF.js. Works on text-based PDFs; not on scans (no OCR).
• Plain text (.txt) — read directly.
• Markdown (.md) — read as text. Headings and lists preserved in the merged output.
• HTML (.html, .htm) — parsed locally, scripts and styles stripped, visible text extracted.
• CSV (.csv) — read as text. Useful for merging data files into a context block.

All extraction happens on your device. There is no server that receives your files, and no telemetry that records filenames or content.

How to use it in 3 steps

1. Add your documents. Drag a folder of PDFs onto the drop area, or click to browse. Up to 30 files at once.
2. Choose your cleanups. Every step is a toggle. Pick a separator style. Click Merge and clean.
3. Copy or download. Pick plain text, Markdown, or HTML. The merged result is one clean file you can paste into ChatGPT, Claude, Gemini, or any other AI tool.

Frequently Asked Questions