Every company formats their receipts differently. An AWS invoice looks nothing like an Uber receipt, which looks nothing like a hotel confirmation email. So how does Folkslip make sense of all of them?
The challenge of receipt parsing
Traditional receipt scanners rely on templates — predefined rules for each provider. This works until a company updates their email format, which happens constantly.
Folkslip takes a different approach. Instead of rigid templates, we use AI models that understand the structure and intent of receipt emails.
What the AI extracts
When Folkslip scans a receipt email, the AI identifies and extracts:
- Provider — who sent the receipt (even if the sender name doesn't match the company name)
- Amount — the total charged, handling different currency formats
- Date — when the transaction occurred
- Invoice/order number — for reference and reconciliation
- Billing period — for subscription receipts
The AI handles edge cases that break template-based systems: receipts in different languages, multi-item orders, partial refunds, and varied date formats.
Privacy by design
Folkslip only reads emails that look like receipts. We don't store raw email content — only the extracted structured data. Your Gmail access is read-only, and you can revoke it at any time.
We also adhere to Google's Limited Use requirements, meaning your email data is never used for advertising, shared with third parties, or accessed beyond what's needed to find and parse receipts.
Accuracy over time
AI parsing isn't perfect on day one, but it improves. As we encounter new receipt formats and edge cases, the system gets better at handling them. The goal is simple: you should never have to manually correct a parsed receipt.
Try it yourself
Connect your Gmail and run a scan. In under a minute, you'll see your receipts extracted and organized — no templates to configure, no rules to set up. Just receipts, ready to go.