Securex: Industry Leader in HR and Social Administration
Securex has long been a prominent player within social administration and human resources, providing comprehensive services to improve workplace environments and operational efficiencies for businesses. Operating across six key business units - Payroll Services, Insurance, SEPP (External Service for Prevention & Protection at Work), Entrepreneurs, MCM, and Consulting - Securex addresses various aspects of business operations. For instance, the Payroll unit ensures accurate and timely payroll processing, while the Prevention department focuses on creating safer and healthier workplace environments. The Lex4You portal (within its Payroll Services unit) offers legal advice, HR news, and essential documents in Dutch, French, and German, supporting businesses in navigating complex regulatory landscapes.
Current Challenges and Opportunities
While Securex is already an established name in the industry, the company recognizes the potential for further efficiency and accuracy improvements. For example, the Prevention department manages extensive advisory documents, and the Legal Knowledge team generates detailed legal summaries and monitors legislative changes. By leveraging AI, Securex aims to enhance processes and reduce manual workload across all business units, ultimately maintaining its competitive edge. This initiative aligns with Securex' commitment to innovation and continuous improvement in service delivery.
The AI Lab Initiative & Securex' Path to AI Integration
The AI Lab at Securex was established to address these operational challenges by organizing and developing PoCs across various business units and domains. This collaborative environment allows for rapid prototyping and testing of AI solutions. Successful PoCs are then evaluated by a project board before being scaled into full production. The AI Lab serves as a fast-moving and autonomous hub for innovation, enabling Securex to experiment with promising use cases without interfering in day-to-day processes and integrate these advanced AI technologies into its operations, improve service delivery, and maintain its leadership in the industry.
Collaborative Process with Faktion
Securex and Faktion began their collaboration by identifying key processes where AI could add the most value and prioritise these based on business value and feasibility on an elaborate AI Roadmap. This involved detailed discussions to understand the specific challenges faced by different departments. The evaluation focused on processes consuming significant employee time and resources, such as translations, tagging historical advisory documents, generating comprehensive advisory reports, retrieving specific pieces of information from large databases, and summarising changes within collective labor agreements for salary calculations.
Formation of the AI Roadmap
The AI Roadmap pinpointed areas where AI could streamline operations and improve accuracy. Securex and Faktion selected use cases based on their potential business value and feasibility of AI implementation. This led to the development of multiple PoCs, each tailored to address specific operational challenges.
Prevention: SEPP Document Tagging, Summarization and Dynamic Keyword Extraction
When writing advisory reports, employees of Securex' Wellbeing and Prevention department (SEPP: Services Externes pour la Prévention et la Protection au travail) often find themselves tasked with sorting through massive document archives to find specific passages in lengthy, historical reports that contain information needed to write a new report. To overcome this problem, Securex began manually tagging and categorising the backlog of thousands of historical advisory documents. The first project aimed to automate this tedious task.
Faktion developed an AI model to automate this process, analyzing document content to assign relevant tags, generate concise summaries, and extract dynamic keywords. This significantly reduced the estimated time and domain expertise involved, improving document retrieval times and allowing employees to focus on higher-value tasks. The combination of a static classification system and a dynamic keyword extraction methodology provides both the robustness and the flexibility for swift information retrieval, for manual queries and for retrieval augmented generation applications.
Customer Care: Supporting Customer Operations with a Knowledge-Based Chatbot
The second PoC involved setting up a chatbot tool for Securex' customer operations employees as they often face the challenge of navigating vast internal knowledge bases to deliver quick and accurate responses to client queries. Their tasks frequently involve searching through multiple databases to find specific client information, responding to identical queries from clients, manually cross-referencing data across different systems, drafting repetitive responses, and verifying the accuracy of data before sharing it with clients. Faktion implemented a Retrieval-Augmented Generation (RAG) system that combines Large Language Model (LLM) capabilities with an advanced information retrieval system and configurable workflow-specific prompt engineering. The RAG Copilot system offers a solution to this manual efforts by leveraging generative AI to assist employees in real time, retrieving relevant information from internal knowledge bases, upon request. This ensures faster, more consistent responses and significantly reduces the need for manual searches, enabling employees to focus on more complex tasks.
Legal: Summarizing Changes in Labor Agreements for Payroll Updates
Payroll Service Providers often face the challenge of staying up to date with frequent legal changes in collective labor agreements, which requires manual adjustments to payroll calculations. Their work involves regularly checking for legal changes affecting payroll, suggesting rule adjustments to maintain compliance, manually updating employee salary details, and preparing client reports that reflect new legal updates. The third pilot automated the summarisation of changes in collective labor agreements, enabling customer care representatives to quickly access concise summaries of the latest changes. For example, when legislation around remuneration changes - such as adjustments in bicycle allowances for commuting - the AI model detects these semantic changes and highlights them, ensuring that only relevant updates are summarised and communicated to the salary engine team. This has significantly improved the efficiency and accuracy of legislative monitoring.
Legal: Conversational AI for Legal Searches in the Lex4You Portal
Legal experts from Securex Lex4You department often spend considerable time manually searching through extensive legal databases to retrieve information for clients, which can delay responses and increase the risk of errors. Their routine tasks include searching for relevant publications, laws, and regulations, cross-referencing past documents, drafting similar legal summaries, addressing frequently asked client questions, and reviewing legislative changes to update internal knowledge. The fourth PoC aimed streamline these efforts, through a copilot that streamlines legal searches and delivers quick and accurate responses, significantly reducing the time experts spend on routine queries. The copilot enhanced the capabilities of the Lex4You portal by seamlessly integrating with Securex' internal databases, providing scalable and accurate legal assistance. It improves document retrieval and client interactions, ultimately elevating service quality and making the process more efficient and scalable for legal teams.
Improving Translation Accuracy and Text Quality
Securex faced challenges with its translation processes, which were either manual or depended on basic machine translation, often leading to inconsistencies in terminology, tone of voice, and industry-specific jargon. This affected the professionalism and efficiency of their services, causing delays and necessitating additional proofreading and correction rounds that slowed down workflows and demanded more resources. Accurate and consistent translations are especially critical for Securex, as their work involves legal and HR documentation where even minor errors can lead to miscommunication or operational issues. Managing large volumes of multilingual content required domain experts to translate, proofread, and adjust text tone to maintain company standards. To address these challenges, in the fifth PoC, Faktion built a genAI tool that not only automates translations, but also optimises text quality. The solution ensures accuracy in acronyms, industry-specific terminology, and tone, resulting in faster, more reliable translations with minimal corrections. This has streamlined Securex’ workflows, enabling them to deliver translations that meet the highest professional standards, while significantly reducing time and effort.
Real-Time Speech-to-Text Transcription and Text Quality Optimisation
Professionals in prevention, wellbeing, and consulting often spend hours transcribing conversations, meetings, and interviews into text for reporting and compliance purposes. This process involves manually transcribing interview recordings, summarizing meetings, formatting reports with detailed notes, proofreading transcription errors, and rewriting content to ensure clarity and adherence to documentation standards.
To address these time-consuming tasks, Faktion implemented a sixth PoC: a generative AI-based speech-to-text service that provides real-time transcriptions integrated directly into their workflows. The solution includes AI-driven enhancements for clearer, error-free documents, significantly accelerating the documentation process. This PoC focuses on improving transcription quality across various business applications by leveraging advanced language models that ensure the accurate use of terminology, acronyms, and names. This enhancement is crucial for maintaining precise records and reducing the manual effort required for transcript corrections, ultimately delivering substantial business value by streamlining operations and improving efficiency.
Conclusion
Securex’ ongoing partnership with Faktion has enabled the integration of AI across multiple facets of its operations, driving significant improvements in efficiency, accuracy, and service quality. Through the AI Lab, Securex continues to innovate and adopt new technologies, ensuring it remains a strong challenger in the competitive market. The successful implementation of these AI POCs positions Securex for sustained growth and operational excellence, reinforcing its reputation as a leader in social administration and human resources.