Institutional Repository of Vilnius University
Klientų aptarnavimui skirti DI pokalbių robotai su žmogaus priežiūra /
Abstract
dc:descriptionThis thesis examines hybrid, humanassisted AI chatbots in customer support. The first part provides an overview of existing chatbot technologies, covering three main classes: rulebased systems, large language model (LLM) based solutions, and hybrid approaches and discusses their respective advantages and drawbacks. Particular attention is paid to the problem of hallucinations and the use of retrievalaugmented generation (RAG) as a possible mitigation strategy. The primary conclusion of this section is that AI alone cannot address all customerservice challenges, making human oversight indispensable; since human time is both limited and costly, optimizing this review process is essential. The growing popularity of the term “augmentation” is also highlighted. The second part describes a practical implementation: an extension of the opensource Kodee platform via a middleware layer that allows a live agent to review, edit, and approve AIgenerated responses in real time. Two Reactbased interfaces—one for the customer and one for the specialist—ensure smooth message exchanges and streamlined agent review. In the experiment, four customersupport specialists completed four tasks both unaided and with AI assistance plus review, achieving an average time saving of approximately 39 percentage points when using the hybrid system. These results confirm that AIhuman augmentation substantially enhances support efficiency.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Klimašauskas, Arnas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD239031176&prefLang=en_US