Izdelava agentnega sistema za podporo zaposlenim v večjih podjetjih

Hladin, Rok (author), Žitnik, Slavko (mentor), Čermelj, Vid (comentor)

Zaposleni v večjih podjetjih pri vsakodnevnem delu potrebujejo hiter dostop do informacij o produktih, storitvah in postopkih. Ker so te informacije razpršene po več virih, je ročno iskanje počasno in pogosto vodi do nepopolnih odgovorov. Ta problem v diplomski nalogi naslovimo z zasnovo, izvedbo in ovrednotenjem agentnega sistema za zaposlene v Telekomu Slovenije, ki temelji na s poizvedovanjem obogatenem generiranju (RAG). Bazo znanja gradi cevovod za zajem podatkov, ki vsebine spletne strani telekom.si pretvori v zapis Markdown, jih razdeli na odlomke in kot vektorske vložitve zapiše v bazo Azure AI Search. Nad njo deluje agentna aplikacija, zgrajena na ogrodju LlamaIndex po vzorcu ReAct, ki ob vprašanju izvaja hibridno iskanje in odgovor podpre z navedbo virov. Zaposleni do sistema dostopajo prek pogovornega vmesnika z enotno prijavo, poleg oblačnega jezikovnega modela pa podpira tudi lokalno gostujoče modele znotraj infrastrukture podjetja. Sistem smo ovrednotili na 105 dejanskih vprašanjih testnih uporabnikov: delež uspešnih odgovorov znaša 94,3 %, prvi del odgovora pa uporabnik praviloma prejme po približno petih sekundah. Prihranek časa potrjuje tudi anketa med testnimi uporabniki. Glavna omejitev in najpomembnejša smer nadaljnjega razvoja ostaja pokritost baze znanja.

• Language: Slovenian
• Keywords: agentni sistem, veliki jezikovni model, s poizvedovanjem obogateno generiranje, vektorska podatkovna baza, hibridno iskanje
• Work type: Bachelor thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2026
• PID: 20.500.12556/RUL-187361
• Publication date in RUL: 10.09.2026

Secondary language

• Language: English
• Title: Development of an Agent System to Support Employees in Large Enterprises

Abstract

Employees of large enterprises need fast access to information about products, services, and procedures. Because this information is scattered across multiple sources, manual searching is slow and often yields incomplete answers. This thesis addresses the problem by designing, implementing, and evaluating an agent system for the employees of Telekom Slovenije, based on retrievalaugmented generation (RAG). Its knowledge base is built by a data-ingestion pipeline that converts the telekom.si website content into Markdown, splits it into chunks, and stores them as vector embeddings in Azure AI Search. On top of it runs an agent application, built on the LlamaIndex framework using the ReAct pattern, which performs hybrid search and cites sources in every answer. Employees access the system through a conversational interface with single sign-on. Alongside a cloud language model, the system also supports locally hosted models within the company infrastructure. We evaluated the system on 105 real user questions: the answer success rate is 94.3%, and the first part of the answer typically appears after about five seconds. A survey among the test users confirms the time savings. Knowledge-base coverage remains the main limitation and priority for future work.

• Keywords: agent system, large language model, retrieval-augmented generation, vector database, hybrid search