20.500.12556/DKUM-98754 Razvoj prototipa ui glasovnega asistenta za podporo recepcijskim procesom Development of a ui voice assistant prototype to support reception processes V diplomskem delu je obravnavan razvoj prototipa umetnointeligenčnega glasovnega asistenta, namenjenega podpori recepcijskim procesom v malih in srednje velikih podjetjih. Namen dela je bil preveriti, ali je mogoče z razpoložljivimi sodobnimi orodji za umetno inteligenco razviti funkcionalen prototip, ki bi prevzel tipične naloge receptorja – sprejemanje klicev, odgovarjanje na vprašanja in rezervacijo terminov – ter pri tem ohranil naravno glasovno komunikacijo v slovenskem jeziku. Za izvedbo smo uporabili metodologijo dizajnerske znanosti, ki je posebej primerna za razvoj in vrednotenje IT-artefaktov. Prototip smo razvili na podlagi ogrodja Next.js 15 v kombinaciji z OpenAI Realtime API, ki je bil v splošno dostopni obliki objavljen oktobra 2024. Za prenos zvoka smo uporabili protokol WebRTC, ki zagotavlja nizko zakasnitev in dvosmerno glasovno komunikacijo v realnem času. Rezultat naloge je delujoč spletni prototip glasovnega asistenta "Ana" za fiktivno podjetje NovaTech Support, d. o. o., nameščen na platformi Vercel in podprt s podatkovno bazo Neon PostgreSQL. Asistent vodi naravne pogovore v slovenščini, zbira podatke za rezervacijo termina, pošilja potrditvena elektronska sporočila prek storitve Resend in samodejno zaključi pogovor. Ugotovili smo, da je razvoj takšnega prototipa realno izvedljiv v razumnem časovnem okviru, ključni preostali oviri za produkcijsko rabo pa sta odsotnost integracije s pravimi poslovnimi sistemi in nedeterministično vedenje jezikovnega modela. This thesis addresses the development of a voice assistant prototype designed to support reception processes in small and medium-sized enterprises. The aim was to verify whether currently available artificial intelligence tools can be used to develop a functional prototype capable of taking over typical receptionist tasks – answering calls, responding to questions, and booking appointments – while maintaining natural voice communication in the Slovenian language. The work was carried out using the Design Science Research methodology. The prototype was built on the Next.js 15 framework in combination with the OpenAI Realtime API, which became generally available in October 2024. The WebRTC protocol was used for audio transmission, providing low latency and bidirectional real-time voice communication. The outcome is a functioning web prototype of a voice assistant named "Ana" for the fictional company NovaTech Support d.o.o., deployed on the Vercel platform and backed by a Neon PostgreSQL serverless database. The assistant conducts natural conversations in Slovenian, collects booking data, sends confirmation emails via the Resend service, and automatically ends the conversation when the caller says goodbye. We found that developing such a prototype is realistically achievable within a reasonable timeframe; the main remaining limitations are the absence of integration with real business systems and the non-deterministic behaviour of the language model. umetna inteligenca glasovni asistent avtomatizacija poslovnih procesov recepcijski procesi prototipiranje artificial intelligence voice assistant business process automatization reception processes prototyping true true false Slovenski jezik Angleški jezik Diplomsko delo/naloga 2026-07-06 14:29:18 2026-08-04 21:37:24 2026-08-05 03:06:15 0000-00-00 00:00:00 2026 0 Maribor 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 286921219 VS_Babic_Mark_2026.pdf VS_Babic_Mark_2026.pdf 1 9DB41DD51C5DBAE2823ADA5DE2EC79C1 a9f1340962528df40ac1380d85691476c321e16938a1cc3065e96eec32b4a34b 9c8c6e9b-79db-11f1-9e97-00155d580e34 https://dk.um.si/Dokument.php?lang=slv&id=207996 VS_Babic_Mark_2026.zip VS_Babic_Mark_2026.zip 1 95FD709806134DDC4ECFDF31606CD010 9d918920acd05f88f58bc3f015f12ac722310ff3b232a858fbfd379b3352d9a3 fd455cfb-7954-11f1-80c7-00155d580e34 https://dk.um.si/Dokument.php?lang=slv&id=207960 Fakulteta za organizacijske vede 0 0 0