Dainty Trading
Home / Projects / Receipt Bridge

Receipt Bridge

Field workers text a photo of a receipt to a shared number. OCR extracts the data automatically and surfaces it in a finance dashboard. No paper, no manual entry, no missed reimbursements.

Category  AI · Field operations Stack  Django 5 · Celery · Postgres LLM  Google Cloud Vision Status  In development

The problem

Field-based teams accumulate paper receipts. The receipts get lost in vehicles, washed in pockets, or photographed and forgotten in a camera roll. Finance ends up chasing them at month-end. Existing “receipt scanning” apps require an app install, an account, and a habit — three things tradies, drivers, and field reps don’t want.

What we’re building

Receipt Bridge gives the team a single SMS number. Workers text a photo of the receipt; that’s the entire UX. Twilio receives, our Django backend kicks off a Celery job that runs Google Cloud Vision to extract vendor, date, total, and line items, and the structured record lands in a finance dashboard for review. Originals go to S3 with retention rules.

The AI angle

Receipt OCR is a solved problem when receipts behave. Real-world receipts — faded thermal paper, oblique angles, glare, half-folded — are not solved. We feed the raw OCR plus the original image into a quick verification step that catches obvious mistakes (totals that don’t add up, vendor names misread as nonsense) before they hit the dashboard.

How it’ll be used

  • Trades businesses with multiple vans on the road.
  • Sales teams with travelling reps and per-diem expenses.
  • Property managers with maintenance contractors submitting reimbursable receipts.

Where we are

SMS ingest, OCR, and the review dashboard are working. Next milestone is exports to common bookkeeping formats (Xero, MYOB, QuickBooks) so finance can clear the queue with a single button.