CyberNet AI has introduced a platform of voice AI agents in Uzbekistan, designed to automate the work of contact centers in banks, retail, and the telecom sector. The system supports Uzbek and Russian languages and is already positioned as a tool capable of reducing call center costs by up to five times compared to traditional operators.

The solution includes three main modules: scenario-based voice robots, a generative AI agent without fixed scripts, and a call analysis system. The first option operates according to predefined algorithms, recording customer responses and converting dialogues into text. The second generates responses in real-time based on the user's request, and the third analyzes conversations, identifying operator errors and service quality.

Special emphasis is placed on its proprietary language model, developed without using ready-made foreign solutions. Turkic languages and real dialogues, including Uzbek, Kazakh, and Kyrgyz, were used for training. Due to the lack of quality data in Uzbek, development took about four years, including testing and refinement stages.

The platform operates both in the cloud via a SaaS model and on local customer servers, with data remaining within the country. In addition to voice calls, the system supports SMS, chats, mobile applications, and web channels, and can also determine the emotional state of the interlocutor – for example, irritation or a raised tone.

Real-world cases have already shown increased efficiency: in the telecom sector, sales conversion for AI agents reached 14–15% compared to 10–11% for live operators. Developers emphasize that the goal of the technology is not to replace people, but to shift routine workload to AI so that employees can focus on more complex tasks.