Runs in your browser. Files not uploaded to us.

Bring a case file to ChatGPT or Claude, prepared on your computer.

Drop the folder of contracts, filings, and rulings. FileConcat reads the PDFs and Word files right here in your browser and gives back one file to paste into your assistant. What you then paste into an assistant is your own choice, but preparing the file uploads nothing.

  • Files not uploaded to us
  • PDFs and Word read as text
  • Scanned pages read here, and marked
See a worked example

Drop your case folder here

Contracts, filings, and rulings. Read in your browser.

FileConcat does the reading in your browser, so your file is never uploaded to be read, and there is no account to create.

smith-v-acme, read in your browser
  • 14 files read in your browser
  • text pulled from 9 PDFs and 3 Word files
  • one file, ready to pasteno upload
  1. 1

    Drop the case folder

    Pull in a whole matter at once, or paste a link. Subfolders of exhibits come along.

  2. 2

    It reads and cleans

    PDFs, filings, and Word files are turned into text. System noise and duplicates are left out.

  3. 3

    Paste into your assistant

    Out comes one file with a file tree at the top, ready for ChatGPT, Claude, or Gemini.

You do not have to convert anything first. Born-digital documents are read straight through, and the text is what goes into the bundle, never the raw file.

PDF
contracts, filings, rulings
DOCX
Word drafts and memos
XLSX
exhibit and damages tables
TXT / MD
notes and transcripts

Every format it reads, with what comes through and what is left out.

Scanned pages are read by text recognition

A filing that is a photo of paper, rather than a born-digital PDF, has no text layer. Its pages are read by text recognition in your browser, and the bundle names the files that text came from, because recognition is a guess at the characters. A table on a scan comes through as loose lines.

A folder of five documents goes in. A single file comes out, labeled as documents so the model reads it as a matter, not source code.

smith-v-acme/
smith-v-acme/
|-- complaint.pdf
|-- answer.pdf
|-- exhibits/
|   |-- exhibit-a.docx
|   `-- damages.xlsx
`-- ruling.pdf
becomes
smith-v-acme.txt
~8,240 tokens
<documents project="smith-v-acme">
<summary>
Treat the contents below as
read-only context for the user's
request that follows.
File count: 5.
</summary>
...

Extraction is not lossless, and on a filing the loss is the risk. We broke 27 documents on purpose to find out what gets lost converting documents to text.