Archive

Problem:

A web-based property agent hired staff to search for, identify, and enter individual property data into their website for potential customers. Having limited resources coupled with the task of finding and identifying thousands of available properties using staff meant the company’s database was limited, resulting in stagnating stales.

Result:

Our technology automatically found relevant data from targeted websites, determined the usefulness of the data and cleaned it according to client’s needs. Saved costs by eliminating need to allocate staff for these processes and increased database size by a factor of 10, leading to a significant rise in property sales.

Problem:

A large Japanese institution wanted to add a coding component into flying drones to make both coding and drone-flying more interest, as well as raise its educational value in the eyes of students.

Result:

We were able to develop an application for pupils to programme, control, and fly drones, allowing primary school students to see the actual physical results of their coding efforts.

Problem:

A courier service had staff manually enter handwritten delivery details on package slips to enable quick route planning and delivery. It took 2 staff members to spend 3 minutes entering data on an individual slip and 2 minutes to verify each delivery for over 1 million deliveries per day.

Result:

Nexus-developed machines recognized different handwriting styles as well as quality of input, guessing the correct information. This made it possible to input required data within the 2-hour window for 1 million+ deliveries per day with the ability to recognise different handwriting styles and quality of input, guessing the correct information.

Problem:

The city of Doha, Qatar needed to track the 20,000+ personnel who moved in and out of the tunnels of the Doha Metro that was being built for safety precautions.

Result:

We created a system to track and monitor personnel inside the tunnels and issue advance warnings of any emergency situation, allowing real-time monitoring of deeper insights into staff movements. The next phase involves extending the capability to predict personnel traffic flows and even accidents.

Problem:

A major Japanese furniture chain wanted to improve online sales by better recommending pieces for customers, but it lacked profound historical data due to furniture not being frequently purchased.

Result:

We trained machines to use images of similar designs to make recommendations within parameters established through limited human preferences identified in a prior survey, increasing the chain’s online sales conversion rate by 35%.

Problem:

When dealing with customer changes of bank account, a leading life insurance company was required to match their clients’ information in their database. To do this, over 100 employees were staffed and spent hours manually processing physical forms and inputting data for the 1,000+ account change requests each day.

Result:

We developed machines to collect the needed information on the physical forms to automate the machine process, enabling the company to use their allocated resources 80% more efficiently, resulting in savings of over USD $2 million per year.

Problem:

A Japanese car manufacturer hired staff to read and match car parts with specific information stored on computer systems prior to sending the data to suppliers and divisions. Staff had to search out information for anywhere between 150-600 parts per car manually.

Result:

We developed machines to automatically read off relevant information and match this information with the correct parts, reducing time and effort from engineers during the extraction phase.

Problem:

A large global bank’s UK sales quality (SQ) team reviewed the sale of financial products to ensure regulatory compliance. Only 10%-15% of sales were actually checked, with a group of over 120 reviewers manually reviewing 10+ data sources each. Each review took roughly 90 minutes to complete.

 

This labour-intensive process left the bank with a higher level of compliance risk, resulting in various faults and errors in unchecked cases going uncorrected. The accuracy levels were also of concern, as inconsistency and potential oversights were made due to human involvement. Lastly, the entire process was slow, with time-consuming reviews and delayed feedback.

Result:

In order to improve the speed of the bank’s review process and exceed the statutory review quota, Nexus proposed to automate this internal process to make checks faster, more comprehensive, and with a higher speed and accuracy by checking 100% of samples and conducting verifications in real-time using AI models to reach near 100% accuracy levels.  All this enabled the elimination of the backlog of checks, ultimately moving them to the point of sale.

 

Nexus successfully helped the global bank standardise and reduce the time spent in their internal sales quality process, check 100% of sales with near-perfect accuracy, free-up their employees’ capacity, and improve customer service.

en_USEN
viVI jaJA en_USEN