匠人学院 JR Academy
You are viewing an example. The people, companies, orders and earnings below are fictional and show the full flow. Buttons here never create orders or charges.Back to AI experts →
Example (fictional)

A complete example

A Sydney customs brokerage, end to end: AI diagnosis → choose an expert → booking → session summary → rebook, plus what the expert sees for orders and earnings.

Orders

Earnings ledger

OrderLearner paidPayment feePlatform feeExpert receives
DEMO-20261104-0815A$210.00− A$4.18− A$42.00A$163.82
This monthA$210.00− A$4.18− A$42.00A$163.82

Amounts use the current fee rules: platform fee 20%, payment fee 1.85% + A$0.30 per order.

On day 1 of each month the previous month's statement is generated: pending → confirmed → paid.

Order detail · DEMO-20261104-0815

Learner's task and AI diagnosisWe are an 8-person customs brokerage in Sydney. Every day we key 40–60 commercial invoices into our customs software by hand and often get unit prices and quantities wrong. We want AI to read the invoices and enter them, with staff only checking the ones that look wrong.
Estimated expert time
3 h
  1. Decide which fields to read

    Go through the customs software entry screen and list the fields to read from each commercial invoice (description, quantity, unit price, currency, country of origin, HS code and so on).

  2. Test accuracy on real invoices

    Run 20 past invoices from different suppliers and mark where AI often misreads, such as handwritten notes and multi-page tables.

  3. Export in a format the software can import

    Turn the results into a CSV the customs software can import, so nobody has to copy and paste.

  4. Design the human review step

    Lines with low confidence or totals that do not add up are highlighted automatically for staff to check before submitting.

Mark complete and write the summary

Required when completing AI services. Edits here stay on this page only and reset on refresh.