In short: Tiro reads the values you need from PDFs – header fields such as invoice number, date and total as well as the individual line items – and presents them as a table. You upload the file, the AI proposes the columns, and you download the result as Excel or CSV. No template, no training, no shifted columns.
Why "convert PDF to Excel" usually goes wrong
A PDF knows nothing about tables. It stores characters with coordinates. A classic converter tries to reconstruct rows and columns from that – and fails as soon as an invoice has multi-line items, shifted blocks or a second page. Everyone who has tried it knows the result: amounts as text with thousands separators, dates Excel cannot sort, and columns named differently on every document.
The actual need is rarely "the layout in Excel" but "these seven values from every document in the same seven columns". That is exactly what Tiro does.
What Tiro does differently
- Fields instead of layout. You decide which values go into the table. On the first upload the AI proposes fitting fields for your document type; afterwards they are editable in the extractor.
- Line items as rows. Invoice positions, bank statement transactions or delivery note articles are output as their own table rows – one row per item, across all documents.
- Normalized values. Dates come as ISO dates, amounts as real numbers. Excel can sum, filter and sort immediately.
- Scans included. Text layer and page image are evaluated together, which is why scanned or photographed documents work too.
- No identical layout required. Every document is read individually. Tiro is strongest with recurring documents: create the extractor once, then upload whatever comes in.
Typical documents
| Document | Typical fields | Line items |
|---|---|---|
| Incoming invoice | Invoice number, date, supplier, net, VAT, gross | Article, quantity, unit price |
| Bank statement | Account number/IBAN, period, opening/closing balance | Booking date, reference, amount |
| Delivery note | Delivery note number, date, recipient | Article number, description, quantity |
| Purchase order | Order number, date, delivery address | Article, quantity, price |
Limits – honestly
Handwritten notes and very poor scans (skewed, blurry, cropped) are hard for any automatic recognition; in such cases Tiro shows the document's status and the recognized values so you can check instead of silently getting wrong numbers. The output is one table per extractor – if you need an exact replica of the page layout in Excel, a layout converter serves you better.
For developers
Everything that works in the browser also works via REST API with an API key: upload a PDF, fetch the result as JSON. Same rules and limits as in the UI.
How it works
- Create an extractor
After signing up you give the extractor a name, for example "Incoming invoices". That is all the setup there is.
- Upload a PDF
On the first upload the AI reads the document and proposes fitting fields – document number, date, amounts, line items. You can adjust, rename or drop them.
- Download the table
Every further PDF lands in the same table structure. One click gives you Excel or CSV with real numbers and ISO dates.
Frequently asked questions
How do I convert a PDF into an Excel table?
With Tiro you create an extractor once, upload the PDF and get the values as a table in the browser. From there it is one click to an Excel or CSV file. For further documents of the same kind, uploading is all it takes.
Does it work with scanned PDFs?
Yes. Tiro reads the PDF's text layer and additionally looks at every page as an image, so scans and photos of documents are processed just like digitally created PDFs.
How is this different from a PDF-to-Excel converter?
A classic converter copies the page layout into cells – columns shift, numbers stay text. Tiro extracts the values you need into a fixed table structure with normalized numbers and dates.
What does it cost?
Getting started is free, 20 pages per month, no credit card. One page equals one credit – that stays the only metric, also in the paid plans to come.
Where are my documents processed?
The app runs on a server in Germany. AI processing happens via AWS Bedrock in the EU region Frankfurt, with no training on your data. Details are in the privacy policy.