As part of Customer Invoicing, a Garage had to enter each part used for a vehicle repair to a job sheet. This mundane process involved copying part numbers, quantities, and unit prices from a PDF supplier invoice, and was being conducted by the Car mechanics themselves. With mechanic recruitment an issue, this was wasting the time of a front-line resource. Accordingly it was decided to automate the process using RPA.
The manual process included:
- Receiving paper invoices.
- Logging into the Garage’s Software Application
- Search for the vehicle registration on the top of the supplier invoice.
- Find the open “job”.
- Add all invoice items to the jobsheet.
- Save it.
- Run separate workflow for stock items
Since the invoices where scanned, and were essentially images, Optical Character Recognition (OCR) was necessary to read the vehicle registration field, and the grid of parts data on the invoice. Accepting that the Human doing the same process would not be 100% accurate, it was still important that a high level of precision was achieved.
A combination of AWS Textract and Google Vision was integrated into the RPA workflow to achieve a high level of accuracy.
- Garage Software Application
- AWS Textract (OCR)
- Google – Gmail, Sheets, Drive and Vision (OCR)
Process Flow Map
RPA Bot Steps
Human scans supplier invoices and uploads to Google Drive
Triggered manually from the Desktop, a summary of the RPA Bot steps looked like this:
- Open Garage Software Application
- Access Google Drive via API, split multiple page PDF’s and move to AWS bucket
- Extract Data from PDF using AWS Textract
- Run Vehicle Registration Field through Google Vision to increase accuracy
- If invoices are Stock items, run through separate Subprocess
- Enter vehicle registration number into search bar
- If registration does not exist, add invoice details to Google Sheet
- Enter part details to Job Sheet
- Send Status email with summary of run to specific recipients
Process as % of Employees Time
Car mechanic capacity increased and additional vehicle revenue possible