AI to Take Over Up to 30% of Procurement Functions in Moscow by 2030
New technologies are being introduced only in areas that do not require legally binding decisions and where bias toward any outcome is not possible.
Head of Moscow’s Department for Competition Policy Kirill Purtov said that by 2030, AI technologies could handle up to 30% of the current workload in the city’s public procurement system. In practice, AI agents will take over routine and analytical functions within an already digitized environment.
This shift appears particularly logical in the capital, where Moscow represents the most mature public procurement market in Russia in terms of digitization. In 2024 alone, the city’s procurement volume reached 1.9 trillion rubles (approximately $21 billion). That means roughly one in every seven rubles in Russia’s procurement system was spent in Moscow. By 2025, the city was already concluding between one and three digital transactions daily. This scale reflects the role of public procurement as a major demand driver for domestic solutions across GovTech, process mining, RPA, big data analytics, AI assistants, scoring systems, risk management tools and integration platforms.
AI is steadily moving into the category of core infrastructure tools for public administration. As a result, procurement for socially critical institutions and sectors is becoming faster and more responsive on the ground.
Scalability and Potential Risks
Other countries are likely to examine Moscow’s experience closely. AI assistants operating at scale could appeal to governments undergoing their own digital transformation. Meanwhile, results in the domestic market already show strong momentum: AI has been deployed to support users of EAIST (Edinaya avtomatizirovannaya informatsionnaya sistema torgov – Unified Automated Information System for Procurement), and specialists report that after its rollout, the number of requests requiring manual handling dropped by more than half.
The next phase will expand AI into more sensitive parts of the procurement cycle, including forecasting, analysis, risk monitoring, anomaly detection and support in drafting technical specifications.
At the same time, these areas introduce significant constraints. AI systems depend heavily on data quality, transparency, human oversight, security and the explainability of decisions. The central challenge is to prevent algorithmic bias in the allocation of public funds.
Toward Robotized Procurement
For that reason, Russia’s Ministry of Finance took a cautious approach as early as 2022, avoiding full automation and retaining human involvement in procurement procedures. Even so, by autumn 2025, the Moscow department had already publicly discussed a transition toward robotized contracting models.
At the same time, Sber launched services for market participants where AI helps analyze documentation, risks and contract conditions. These tools are now used not only in government systems but also in the private sector.
Similar trends are visible globally. AI technologies are already being applied to analyze spending, manage risks, identify suppliers and automate operations. Rapid adoption in public administration is also driving stricter requirements for transparency and oversight. In the European Union, the AI Act has already established a regulatory framework for AI systems used in the public sector and administrative processes.
AI Focused on Low-Risk Areas
Given these conditions, AI adoption is expected to grow first in lower-risk areas, including user support, data search and classification, preliminary document analysis, demand forecasting, scoring and anomaly detection. Over time, it will expand into cases involving legally significant decisions, disputed outcomes and higher levels of responsibility. Demand will be strongest for applied solutions tailored to specific sectors.
This transition is expected to benefit all stakeholders. Government agencies will shorten procurement cycles and reduce routine workloads for specialists, while city services will become faster and more predictable for everyday users.
