AI Goes to Work in Government
Generative AI is set to expand across every division of the Russian government's executive office following a six-month pilot program. More than 130 employees from eight departments tested four domestic AI services – Yandex's Alice AI, AI Studio, and Neyroyurist (AI Lawyer), along with Sber's GigaChat. By the end of the trial, roughly 80% of participants had become active users.
The AI models were used to draft meeting minutes from transcripts, prepare briefing materials, compare different versions of documents, route citizen requests, and create first drafts of press releases and analytical reports. The pilot operated in two separate environments: an open environment for processing public documents and information from publicly available sources, and a closed environment with no internet access. The next planned step is to integrate AI into the government's electronic document management system.
Regions to Get Ready
Domestic AI models are becoming a standard part of day-to-day operations across key government institutions. That trajectory aligns with the goals of the national project Data Economy and the Digital Transformation of the State. The solutions tested during the pilot may eventually be deployed through the Upravleniye (Management) State Automated Information System and made available to federal agencies and regional governments, including at no cost. That could be particularly valuable for regions that lack large IT teams, sufficient computing resources, or the funding needed to build their own AI systems.
Naturally, challenges remain. AI models can still generate factual errors or fabricated information, and data protection continues to require careful oversight. Government employees also need training, while responsibility for final documents and official decisions must remain clearly assigned. Even so, these issues are expected to be addressed over time, and some of the tested products could eventually find markets beyond Russia, particularly in the Eurasian Economic Union (EAEU), the Commonwealth of Independent States (CIS), and other countries building digital government platforms independent of Western cloud infrastructure.
A Systematic Rollout
The use of AI in regional government began as early as 2023. At that stage, machine learning supported the processing of citizen requests, urban environment analysis, transportation monitoring, and forecasting of social and economic indicators.
In 2025, the Russian government established an AI Development Center. That same year, AI began supporting the analysis of government program implementation by forecasting potential schedule delays and assessing overall progress. By 2026, AI training for civil servants had become systematic, with programs focused on practical mastery of the tools already being deployed across government agencies. Regional AI platforms have also emerged, including GosPrompt, which supports approximately 150 government workflows.
More AI Models Ahead
For the foreseeable future, AI systems are expected to remain decision-support tools rather than decision-makers. They will assist with information processing, while meaningful government decisions will continue to be supervised by public officials. Their long-term impact can only be measured once clear performance indicators are established, such as faster response times, fewer errors, and measurable time savings. Ultimately, however, the most important benchmark will be whether citizens see tangible improvements in the quality of public administration.
For Russia's technology sector, broader AI adoption across government could also create demand for additional domestic products. That includes secure computing platforms, data management systems, cybersecurity solutions, and specialized language models designed for public-sector use.
