Success
Stories

As the client’s user base rapidly expanded, they needed a scalable solution to manage a high volume of customer inquiries efficiently, particularly around personal investment topics.

The client required a virtual assistant within their mobile application to provide automated consultations on personal investments, aiming to reduce dependency on a large contact center.

We developed and implemented a virtual assistant into the mobile app, specifically tailored to handle brokerage-related inquiries. Within three months, the assistant was trained in brokerage terminology, equipped with over 1,500 unique communication scenarios to ensure accurate and relevant responses.

The virtual assistant now successfully processes over 80% of user requests, effectively replacing a contact center that would otherwise require 50 to 100 operators. With over 200,000 conversations managed each month, the solution has significantly reduced operational costs and improved response times for users.

The client’s contact center robot handled a substantial workload, managing 600,000 user requests per month but accurately addressing only 60% of inquiries. They sought to increase automation and improve the accuracy of responses to reduce strain on human agents.

The existing contact center robot was processing a high volume of requests but required improvement, as only 60% of user questions were handled accurately. Enhancing the accuracy and coverage of automated responses would significantly reduce the contact center’s operational load.

We developed a new model trained on extensive historical data from the contact center’s correspondence logs. Within one month, we compiled a comprehensive knowledge base tailored to common inquiries and client-specific details. This enriched knowledge base was then integrated with the contact center robot to enhance its response accuracy.

Following implementation, the proportion of accurate, automated responses increased from 60% to 72%, reducing the overall load on the contact center by 25%. This improvement enabled more efficient operations and allowed human agents to focus on complex or high-value inquiries.

The client, a financial services firm, needed a solution to streamline the processing of Financial Offering Memorandums (FOM) to enhance the efficiency of their analysts in reviewing and analyzing document content.

Processing financial PDF documents was a time-consuming task, requiring analysts to manually examine and compare clauses across documents. The client sought a solution to automate this process, allowing analysts to handle a higher volume of documents with improved accuracy.

We developed the FOM-Processing System, an advanced Natural Language Processing (NLP) tool that processes financial PDF documents. This system enables users to upload a PDF, automatically extract and analyze all clauses, and compare them against clusters defined by financial experts. Each clause is assigned a score, alongside an aggregate score for overall document evaluation.

The FOM-Processing System significantly accelerates document processing times, enabling financial analysts to review documents much faster than before. This improvement not only increases productivity but also enhances the accuracy and consistency of document evaluations.

The client required an advanced solution to monitor road infrastructure and identify defects through video footage, aiming to improve maintenance efficiency and prioritize repairs based on real-time data.

Detecting road infrastructure issues manually is labor-intensive and prone to inaccuracies. The client needed an automated system that could accurately identify and categorize road defects, integrate geolocation data, and streamline reporting for maintenance teams.

We collaborated with the client to design a comprehensive AI-powered solution using convolutional neural networks (YOLO) tailored for infrastructure analysis. Key steps included:

• Developing the software concept, project roadmap, and detailed technical specifications.
• Creating a data collection methodology optimized for video analysis.
• Training an AI model to detect road infrastructure elements and identify defects directly from video footage.
• Building a geoinformation service to map identified infrastructure damage.
• Implementing an automated reporting feature to inform maintenance teams of prioritized repair needs.

The solution enables efficient, accurate detection of infrastructure defects, mapped directly onto a geospatial interface for easy visualization. The reporting feature further ensures that maintenance teams receive timely updates, helping streamline operations and reduce repair response times.

The client sought an adaptable solution to automate the extraction of key entities and relationships within complex documents, specifically tailored for the fields of law and medicine, where accuracy and relevance are paramount.

Extracting named entities and identifying relationships within text manually is time-intensive and requires high accuracy, especially in specialized fields like law and medicine. The client needed a versatile system that could handle these requirements across different domains.

We developed a universal core system capable of automatically extracting named entities and identifying relationships between them. The system was then customized to meet the unique requirements of legal and medical texts, ensuring precision and relevance in each field.

The implementation of this universal core significantly reduced data processing time and improved the accuracy of information extraction. By automating the text analysis process, the client achieved higher processing efficiency and an enhanced quality of results, supporting more reliable decision-making in both legal and medical applications.

The client had an extensive archive of chat conversations with customers but lacked a clear, actionable understanding of the key insights within this data. They sought a way to analyze and interpret this information to improve customer service, identify product issues, and predict customer behavior.

With a vast volume of unstructured conversation data, the client needed a solution to systematically analyze customer interactions, identify common themes, detect service quality trends, and recognize indicators for customer purchase likelihood or churn risk.

We applied advanced vectorization and clustering techniques to structure and analyze the archived conversations. This approach enabled us to extract meaningful insights and segment key information into actionable categories, including:

• The most popular topics of discussion,
• Common issues customers experience with the company’s products and services,
• Insights on customer service quality,
• Predictive signals for customer purchasing intent or likelihood to churn.

This data-driven analysis provided the client with a clear, comprehensive view of customer interactions, empowering them to address common issues, improve service quality, and proactively manage customer retention and sales opportunities.

Our client, Translink—the company managing the OV-Chipkaart system—alongside Rotterdam’s public transport operator RET, aimed to introduce a seamless payment experience, enabling passengers to tap in and out using their contactless bank cards.

The objective was to modernize payment systems to accept contactless bank cards from multiple Dutch banks, requiring a smooth integration with existing infrastructure and reliable performance to ensure ease of use for passengers.

We integrated RET’s metro gates and Translink’sinternal systems with newly developed components to support contactless payments. Additionally, we created a new customer-facing website and mobile app to guide passengers through the new payment options and provide relevant support.

The project successfully enabled contactless payment across the RET metro system, with test cards from all Dutch banks that offer contactless payment functionality being accepted seamlessly. This upgrade simplified the travel experience for passengers and marked a significant step forward in modernizing payment options within Dutch public transit.

Unilode Aviation Solutions, a leading Swiss air carrier, manages a vast fleet of air containers and required a comprehensive tool for real-time monitoring and management of container parameters, including location, temperature, and humidity.

The client needed a robust system to track key metrics for each container in real-time, ensuring efficiency and safety in container management. This required a scalable solution capable of handling high volumes of data and accessible across platforms.

We designed and developed a universal IoT platform and built a custom software solution, FAST, tailored to address all aspects of container monitoring and management. Key features include:

A high-performance backend that processes over 10,000 requests per second, ensuring real-time data handling.
A flexible, cross-platform web frontend to provide accessibility and ease of use.
A mobile application for on-the-go scanning and management of containers.

The FAST platform has streamlined container monitoring and management, providing Unilode with a powerful, real-time tracking tool that enhances operational efficiency and container oversight. This solution has significantly improved Unilode’s ability to ensure container integrity and optimize resource allocation.