The work, written up

Papers from the Infobip Research team – on conversational AI, voice, fairness, and the hard engineering problems inside communication at scale.

Towards improving visual chatbot builders with intent-based Copilots

E. Lacic, P. Fribert, A. Lukac, I. Lovric | March 2025

Intent-based AI copilots can dramatically speed up chatbot creation while cutting errors – automatically generating dialog components from plain text inputs. Lightweight models keep up with much larger ones, making this practical for real-world platforms.

On the language and gender biases in PSTN, VoIP and neural audio codecs

K. Altwlkany, A. Kuric, E. Lacic | June 2025

Analysing over two million speech samples, we found systematic bias patterns in widely used audio codecs – neural codecs favour certain language groups, PSTN codecs give better quality to male voices. Audio compression has a direct, measurable impact on fairness.

Vocalize: Lead acquisition and user engagement through gamified voice competitions

E. Teskeredzic, M. Paric, A. Sestic, P. Fribert, A. Lukac, H. Hadzic, K. Altwlkany, E. Lacic | September 2025

We built Vocalize – a voice-driven gamification platform that helps brands run interactive audio competitions. Speech processing + generative AI + four large-scale event deployments. High participation rates, strong user retention, and measurable lead conversion across all events.