Artificial Intelligence LLM for smarter access to documentation – Public Management Summary
Worldline, a leading provider of payment and transactional services, offers a wide range of solutions to its clients, each tailored to specific countries, purposes, and tools. To seamlessly integrate these solutions with other hardware and software, Worldline relies on extensive documentation.
Gafirov, Alexander & Zoller, Anne-Sophie & Stoll, Fabian & Selvakumar, Jayalakshmi & Teklit, Dejen, 2024
Type of Thesis Projektarbeit/Praxisprojekt
Client Worldline
Supervisor Wache, Holger
Views: 192 - Downloads: 72
However, the sheer volume and complexity of these manuals often make it challenging for software engineers to locate the information they need quickly and efficiently. This frequent need to seek assistance from the Worldline team is time-consuming, inefficient, and ultimately costly for both parties.
Recognizing the need for a more streamlined and user-friendly approach, Worldline partnered with the FHNW to conduct a practical student project aimed at developing a solution that would empower software engineers to find the information they need within the documentation.
The students embarked on a comprehensive research process, conducting a kick-off meeting, a full-day workshop at the client's premises, and engaging in in-depth interviews with key Worldline personnel. This analysis revealed the challenges faced by software engineers and the potential benefits of utilizing advanced AI technologies to enhance document search capabilities.
The new and rapidly growing landscape of large language models (LLMs) posed a significant challenge in identifying a solution that would truly transform the document search process for software engineers.
After conducting thorough research and evaluating various options, the students discovered amberSearch, a German company specializing in LLM-powered indexing and search capabilities.
AmberSearch's indexing ability enables rapid and accurate text retrieval, akin to a Google search experience. Moreover, its beta version introduces an innovative feature: providing text-based answers analogous to ChatGPT. This advancement empowers software engineers to receive comprehensive and tailored responses to their queries, instead of simply links to relevant sections of the documentation.
The integration of these two advanced approaches – LLM-driven indexing and ChatGPT-style answers – holds immense potential to revolutionize the document search experience for software engineers. By providing direct access to highly relevant information, the solution will significantly reduce the need for frequent support inquiries. This streamlined process will not only save Worldline and software engineers valuable time and resources but could even lead to lower fees for software engineers, as they are faster and further optimizing the overall cost structure.
Studyprogram: Business Information Technology (Bachelor)
Keywords LLM, AI, Chatbot
Confidentiality: öffentlich