Minimising manual effort with advanced cloud analytics

Supporting airline engineers

In the beginning, SoftwareOne worked with the airline’s engineering department. The workers had to spend a lot of time finding relevant information related to aircraft maintenance and management. The goals of the first engagement included:

• Speeding up document processing: data entry and extraction needed to be faster.
• Reducing the number of errors: repetitive and monotonous tasks had to be eliminated, so employees could focus on rectifying issues.
• Standardisation of processes: all documents were to undergo the same automated processing to minimise manual effort and discrepancies it might cause.

What is modern knowledge management?

The concept involves combining knowledge from multiple sources, external (world knowledge) and internal (workplace insights, people’s expertise). Technology now provides the opportunity to aggregate this data and make it searchable using AI and machine learning models. The goal is to bring the right information to the appropriate context fast – regardless of how it’s stored or presented. Cloud-based knowledge management mechanisms allow employees to find the necessary answers significantly faster than before.

Speeding up operations with intelligent search

The airline gets lengthy documents, amounting to hundreds of pages, detailing the technical specifications of their aircraft. These publications are often updated and need to be cross-referenced to identify changes. If an engineer gets a report of a defect to fix, finding a solution manually can be difficult.

To add to the problem, a lot of information was amassed over time, not just in print but also in written form. A standard search would make finding relevant solutions to a problem virtually impossible.

Moreover, because all the information was held in multiple sources and formats, engineers were unable to identify if there were any links or patterns between them. They wanted to minimise this tedious effort, so they could focus on solving potential issues. Azure Cloud provided a solution.

Within 3 months, a knowledge mining mechanism based on Azure Cognitive Search was implemented. It identifies and explores relevant content on a large scale and is fully customisable.

Azure Cognitive Search is available in multiple languages and can work with several data types, like Office documents, PDFs, images, or JSON files.

SoftwareOne also proposed a tailor-made web app interface that employees can operate with ease. Documents are retrieved in order of relevance to the user, speeding up the process. They can also be categorised and filtered by specific fix types, making finding the right answer simpler. Engineers can now find the necessary information in minutes instead of hours.

What is knowledge mining?

Knowledge mining is a process for finding actionable information among different file types, such as documents or images. Powered by AI, it can identify hidden patterns in large volumes of data. It can be used, e.g., for finding specific information among contracts, emails, or PDFs, even in multiple languages.

The service is complemented by the Content Extractor. It uses text mining and machine learning algorithms to locate specified and relevant elements among the entire content.

Enhancing finance operations with advanced analytics

The company wanted to boost its fraud detection capabilities. The key goals of the project included:

• Capture errors and anomalies in outgoing payments, so any potential issues could be resolved sooner.
• Prevent the organisation from transferring unjustified or unauthorised sums of money to minimise the risk of erroneous transactions.

Previously, employees reviewed all transactions manually for potential mistakes, such as duplicates or overpayments. This approach was time-consuming and subject to human error. To ensure no anomalous transactions are missed, the company decided to use analytics services based on the Microsoft Azure cloud to automate the procedure.

SoftwareOne worked alongside the client’s team on creating a solution that would maintain the accuracy of their financial operations. The data was gathered from on-premises sources and moved to the Azure cloud for analysis and storage.

With automated anomaly detection, the client can get information on anomalous transactions in real time, rather than every quarter. This allows them to react and resolve potential issues with payments more quickly. The solution provided a significant ROI within 6 months of operation.

Results are displayed on interactive Power BI dashboards, immediately indicating if there are any causes for concern. Additionally, the machine learning model is able to detect problems that might have been missed in manual detection. Based on continuous feedback, the model will also improve its capabilities over time.

Automating this task allowed the company to save around 600 hours of manual work per year. Instead of gathering and checking data, employees can instead focus on issues brought to light by automated checks.