Last updated: 24 October 2026
ExtractKit extracts structured data from invoices and receipts. Those documents routinely contain names, addresses, tax numbers and bank details, so this page states exactly what happens to them. If anything here is unclear, mail us.
All of it lives in a single SQLite database on our server.
The contents of your documents, and the JSON we extract from them. Uploads are read into memory, processed, and discarded. Nothing is written to disk and nothing is written to the database. PDFs are converted to an image in memory — no temporary files are left behind.
To extract the data, the first page of your document is rendered as an image and sent to:
api.openrouter.ai), acting as the API
gateway, which routes the request to the model provider serving
google/gemini-2.5-flash.
Those providers process the image under their own terms and policies. If your documents contain personal or financial information, that information is visible to them for the duration of the request. We do not send your document to anyone else, and we do not use your documents to train anything.
Our application server is a virtual machine hosted by Linode (Akamai Connected Cloud) in Fremont, California, USA. The database described above lives on that same machine.
The server keeps standard access logs — your IP address, the request path and the response status — plus error messages when something fails. Document contents are not intentionally written to logs.
Usage records are kept for as long as your account exists. Ask us to delete your account and we will remove your email address, your API key and your usage records.
Questions, deletion requests, or anything else: onyx@cntrcd.com