Our Projects

These are some of the solutions and products we developed for our customers

It spans the technology list in our technology section

Automatic Ordering From Paper PO to Fulfilment

Customers send paper copies of their purchase orders (POs). Using optical character recognition (OCR), these POs are converted to XML files and sent to the PO product matching stage. The product matching stage helps identify the best product match from the internal product catalog using NLP technologies and advanced search algorithms. The user confirms the automatic matching through a portal and approves it. Then the final approved XML PO is sent to the ERP for fulfillment. The system learns from the above cycle and improves its future matching for similar items. Product catalog data cleansing and token normalization is an essential step before the search is implemented effectively. The system architecture is closely coupled with endpoints communicating over a proprietary service bus.

Technologies - ABBYY Flexi Capture - Java - My SQL - PHP - Python for NLP - AWS Cloud





Automatic Receipt Digitization

The project uses OCR to convert scanned receipts to the XML data file and sends it as a message to the ERP. The receipts come from multi countries and multi-languages. It is intended for employees on business travel. It helps them claim their expenses with the minimum data entry possible and attach a copy of the receipt. The project requires manual intervention, at the back end, for some of the receipts data with poor quality. A superuser can train the system on new receipts layouts. The system can be extended to reject non-business approved expenses or to reject expense claims from unapproved merchants.

Technologies - ABBYY Flexi Capture Golang windows service to send the XML to the ERP endpoint 

B2B marketplace revitalization

Our customer had a 10 years old marketplace and it is based on old technologies. Our customer wants to capitalize on its success and customer base by taking its marketplace to the next level. Initially, the system needs to be upgraded. Then, the marketplace should support more suppliers and customers and provide more detailed views of data. Moreover, the system should be able to handle larger volumes of transactions. Positive facts are: 
- Both marketplace customers and suppliers are willing to use the newly renovated marketplace and cannot wait for it.
- New customers and suppliers are willing to join the new marketplace business model. It fills a market gap.

Our customer wants to control his investment. So, Kiwi Systems' solution is to revitalize the old system to be ready for modern use. Then when the marketplace creates momentum and cash flow starts to be positive and accumulate, the actual large transformation project will be justified. We managed this by
- Building a proper source control of the existing system
- Upgrading OS and DB to newer versions supported by technology vendors
- Integrating with a payment gateway
- Change the business model to avail of new functionalities to new levels of users.
- Adding new functions to reduce operations overhead and reduce its operating expenses.

The most important value of Kiwi Systems work was
- Building system knowledge which helps support and maintain the system as the demand arises
- Spent fraction of the cost that would be spent in replacing the system. This has availed budget for other business activities to grow our customer's profit 

Technologies - PHP - MySQL - Centos - IBM Cloud 

eCommerce EDI Messaging System Re-Write

It is a real-time and mission-critical service bus. It is used for exchanging trade transactions in different formats including EDI. It works to exchange trade documents/ Messages among different branded systems (Oracle business suite, SAP, Dynamics,....) as well as customized proprietary systems and ERPs. Our mission was to upgrade the infrastructure of the system, make it highly available, change the messaging platform, and re-write the original C++ code with a new up to date technology. The system re-written with the care of each line of code and is reviewed by our top engineers who have 20+ years of software development experience. The system is used by blue-chip customers.

Technologies - Kotlin - Centos 7+ - MySQL - Active MQ

Sales Agent Incentives

On a daily bases, sales data are fed to our Sales Incentives system from the back end system. Store sales agents can see their sales achievements daily. They can check their personal dashboard, store dashboard, and region KPIs. Regional managers can set monthly, quarterly, and yearly sales targets for their teams. Sales targets are linked to incentives. Sales agents compete with each other on sales incentives and monitor daily progress. Sales managers can create sales targets based on company objectives. The system was used by one of the mobile operators in NZ.

Technologies - .Net and C# - SQL Server - Azure Active Directory B2C

iOS Wearable Device Data Capturing

It is an iOS application to capture data from an Empatica device. Data is sent to a Node.js endpoint in the back end server. The data read are human body vital signs. Some of this data is read several times a second and produces around 3M byte per hour. The data transfer rate is large and the system has to keep running in real-time. A real-time app on iOS is never easy as iOS usually kills any process that consumes the CPU for more than a couple of seconds.

Technologies - Swift - Realm DB - Node.js for the back end - My SQL DB - Empatica SDK