A Multi-Tenant Invoice payment solution that uses OCR and multiple payment methods

The customer wanted to build a product that would allow users to take a picture of an invoice, send it over WhatsApp, Slack or email. The product would use an OCR system coupled with a machine learning model to identify the attributes of an invoice including the IBAN, amount, due date, currency and the recipient details. Reminders needed to be sent to customers when an invoice was nearning it's due date. Additional features included scheduling payments, conversations chat via WhatsApp, support for payment methods such as cards & open banking. A rich payment widget that can be embedded on any website to collect payments was also required.

Ongoing

React

AWS EC2

AWS Fargate

PostgreSQL

Go Lang

Slack API

Sendgrid

Chatbots

WhatsApp API

Challenges

A machine learning model had to be built and trained so that it identified all the attributes required to pay an invoice.
Customers could pay invoices through different payment methods including cards and open banking.
Customers can pay an invoice in their default currency which could be different from the currency in the invoice.
Third party stores such as e-commerce stores, utility companies etc. could use the widget on their website as a link and allow users to pay their dues using the product.

Solution

Wednesday setup a team of engineers, designers and a project manager to work closely with the customer.
The team created a product roadmap and milestones with production ready deployments.
Design was central to ensure the user has a seamless experience and paying invoices was as easy as sending a message over WhatsApp, Email or Slack.
The API was built using a multi-tenant micro service architecture. This allowed the customer to be able to deploy the solution in different geographies each supporting a different payment partner.