Document and invoice data extraction
Read invoices, receipts and other documents into a checked spreadsheet, so nobody retypes them.
Being built as a self-serve tool. Register your interest to hear when it is ready.
What it does
You upload PDFs or photos and choose the ones to read. Each document is read on its own: the tool copies what is printed (supplier, number, dates, line items, totals and any extra details you ask for) without correcting anything. Fixed rules then check the figures and dates and mark each document Ready to use or Check before using, with the reasons. You review, correct anything, and download a spreadsheet. Up to 20 documents per run.
Who it is for
- Accountants and bookkeepers handling clients' invoices and receipts
- Small businesses that retype supplier invoices, delivery notes or forms
- Anyone who needs the same few details taken from a pile of documents
Quick start
- 01Upload your documentsOn the tool page, upload PDFs or photos (PNG, JPEG or WebP), or choose files you have already uploaded. Tick the ones to read.
- 02Say what they areChoose invoices, bills and receipts for the full set of invoice checks, or other documents and list the details you want taken from them (for example PO number, delivery date).
- 03Add notes if they helpFor example: we are the customer on these invoices, or dates are day/month/year.
- 04Run, review and downloadStart with the documents marked Check before using: each lists what to look at. Correct values in Edit this result, then download the spreadsheet.
What you give it
- Documents to readRequired, up to 20 files
- The uploaded PDFs or photos to read. Up to 20 at a time, up to 10 pages each. Upload more right on the tool page; new uploads are ticked for you.
- What are they?Required, choose one of 2
- Whether these are invoices, bills and receipts (the tool checks them as accounting documents) or other documents (it reads only the details you list).
- Other details to pick outOptional, up to 15 lines
- Other details to pick out, one per line, such as PO number or project code. For other documents, list every detail you need: it returns one value for each, or leaves it empty and flags it.
- Anything the reader should knowOptional, up to 1,000 characters
- Anything the reader should know, such as which party you are on an invoice or the date format used.
What you get
- A summary
- How many documents were read, how many are ready to use and how many need checking, with totals per currency.
- Each document
- Ready to use or Check before using with the reasons, then who issued it, to whom, number, dates, currency, subtotal, tax, total, line items and your extra details.
- Spreadsheets (CSV)
- One for documents and one for line items, ready to open in a spreadsheet program. Values that a spreadsheet could run as a formula are shown as plain text.
- How we checked these
- The rules used, so you know what a Ready to use mark does and does not mean.
You can edit the result, then copy it, print it or download it as Word, Markdown, plain text or data. Every run is kept in your run history until you delete it.
An example
An example with made-up details: a supplier invoice from a trade supplier.
You give
- Documents to read
- INV-2041 Northgate Supplies.pdf
- What are they?
- Invoices, bills and receipts
You get
- Issued by Northgate Supplies Ltd, number INV-2041, dated 2026-09-02, due 2026-10-02, currency GBP
- Three line items with quantities, unit prices and amounts, a subtotal, VAT and total
- A mark of Ready to use because the line items add up to the subtotal and the subtotal plus tax equals the total
- A row for the invoice and a row per line item in the downloads
What the AI does
It does
- Reads the document, whether a PDF or a photo, and copies what is printed
- Leaves anything it cannot find empty rather than guessing
- Notes anything hard to read
- Ignores instructions printed inside a document
It does not
- Correct a mistake on the document: if the figures do not add up it reports them as printed and flags the document
- Read a document that is blank, too blurred or password-protected
- Write into your accounting software: this tool gives you a checked spreadsheet and does not connect to your systems
- Know whether an invoice is genuine or should be paid
Please check itReady to use means the figures are consistent with each other, not that they are right: a wrong figure printed consistently still passes. Check the documents against anything that matters before you pay or file.
Limits
- Documents to read: up to 20 files.
- What are they?: choose one of 2.
- Other details to pick out: up to 15 lines.
- Anything the reader should know: up to 1,000 characters.
- PDFs can be up to 15 MB and photos up to 5 MB each.
- A document with more than 10 pages must be split into separate files first.
- Your workspace's monthly cap: a run that would take you over it pauses, and carries on where it stopped when you raise the cap.
Your data
- What you enter, the files you choose and the result are stored encrypted with your workspace's own key. Only people in your workspace can open them.
- To produce the result, what you give the tool (and the parts of your Business profile it uses) is sent to our AI provider, Anthropic. It uses it only to answer, and its terms do not allow it to train its models on it. We do not train on it either.
- Files you upload are deleted after 30 days unless you choose to keep them. You can delete a run at any time: what you entered, the result and the progress notes go; only the usage record stays, because it is part of your billing.
- Our service is hosted in the United States.
- Documents are sent to the AI provider as the file itself, one at a time, to be read. They are not kept by us longer than your upload retention allows.
The full details are in our privacy notice.
Questions
What does a run cost?
You pay for the AI it uses, at cost plus a markup, never more than the monthly cap you set. The tool shows the most a run can cost before you start it. The one-time licence fee and the monthly fee for the tool are announced at launch.
Can I change the result?
Yes. Choose Edit this result to change any wording or figure, and Restore the original result to go back. Copy, print and downloads use your edited version.
Is it ready to use now?
Not yet: it is coming soon. Follow it in your account to be told when it opens.
What does Check before using mean?
A fixed rule found something to look at: for example the line items do not add up to the subtotal, a required detail is missing, a date is not a real date, or two documents look like duplicates. The reasons are listed under To check.
What does it cost per document?
Reading a document uses the AI once, so the cost scales with how many you read. The tool shows the most each document can cost before you run it.
Can it write the data into Xero or QuickBooks?
Not today. This tool produces a checked review screen and CSV downloads. Connecting to accounting software is a possible later step, and nothing would ever be written to your system without your approval.
Are my documents used to train anything?
No. See the data handling section below.
If something goes wrong
- A document says it could not be read.
- Upload a clearer scan or a straight-on photo, with good light, and run it again.
- “A file is no longer stored”.
- Uploads are kept for 30 days unless you choose to keep them. Upload the document again and start a new run.
- A PDF has too many pages (more than 10).
- Split it into separate files, one per document, and upload them.
- Every document says Check before using.
- Open one and read the reasons. A common cause is a missing currency or a date in an unusual format: add a note such as dates are day/month/year and run again.
- The same invoice appears twice.
- Two documents with the same issuer, number and total are flagged as duplicates. Remove one from the choice and run again.
Still stuck? Use the enquiry form and say which tool and what you saw.