This is the public demo. Extraction here returns sample receipt data, and your file is not sent to any model. On your own install, the model you configure reads your file.
What happens next
Uploaded receipts start as pending. Back on the report, click Run extraction to read them all with your configured AI model — vendor, date, amount, tax, payment method, and category are filled in for each. You then review and correct anything before export.
Your key, your data
Expensa calls only the LLM provider you configure. Your API key stays in
your own database, and receipts are stored in data/uploads/, protected from direct web access.