AI Speeds Up Moscow Hotline Responses by Up to 40%
An LLM-based voice assistant uses verified knowledge bases and understands conversational speech
The deployment of large language models (LLMs) across key city hotlines in Moscow has accelerated response times to residents’ inquiries by 30–40%, the city’s Department of Information Technology said.
The technology was first introduced in the unified technical support service for the Supplier Portal, and later expanded to the Moscow Government’s general information service, magistrate court hotlines, and the unified dispatch center.
Where the voice assistant previously read out long articles during calls, the upgraded system—based on a domestically developed LLM—now supports more natural conversations. It can interpret context and tone, quickly retrieve relevant information from its knowledge base, and deliver concise answers.
Clear Speech and Context Awareness
The AI assistant can handle multiple questions within a single interaction, helping reduce repeat inquiries by 13%. The system relies exclusively on verified data sources. The combined knowledge base across four hotlines covers more than 600 topics. A dedicated glossary of 1,500 terms with definitions and correct pronunciation helps the system accurately handle complex language.
