Russian regulators are preparing the groundwork for the systemic implementation of AI agents in industrial software. The Ministry of Digital Development and the Ministry of Industry and Trade are jointly developing standards that should streamline the process of integrating these autonomous systems. The key goal of the initiative is to prevent situations where developers block access to their software for third-party agents, thereby limiting the development of sovereign artificial intelligence.
According to my data, the focus is on creating a list of open protocols through which AI agents will be able to interact with various software systems. This document is expected to be presented to the government by December 1, 2026. The concept of an AI agent itself is interpreted as an autonomous software system capable of acting independently in a digital or physical environment to achieve set goals.
The problem of closed software and the maturity of solutions
The most mature agent implementations today are observed in customer service, software development, the financial sector, retail, and telecommunications. In industry, such programs are already helping to work with regulatory documentation, manage procurement, and maintain internal systems. However, as market participants emphasize, manufacturers often close their software to external agents, forcing clients to use only their own solutions. This creates barriers to technological progress and makes the ecosystem fragmented.
Notably, the share of solutions with real agent functionality in the Russian software registry is currently estimated at about 5% of more than 32,300 registered products. For comparison, globally this figure among all AI programs reaches 39%. This is a colossal gap that indicates an early stage of market development in Russia. Analysts estimate the volume of the corporate AI agent market in 2026 at 30 billion rubles, which underscores its significant potential.
The economics and future of agents
Developing and operating software with built-in agents is more expensive, but the savings in working time through automation, according to experts, more than offset these costs. The ultimate goal of any agent is to take over some of the functions of a regular employee, for example, independently sending commands to equipment.
My analysis: The initiative looks timely and necessary. Without unified standards and open protocols, the market risks turning into a set of isolated solutions, which will slow down technological development. However, success will depend on how strict the requirements for developers turn out to be. If the standards remain advisory, major vendors will continue to ignore integration, and then we will not see a significant increase in the share of agent-based solutions in the registry. Time will tell whether the state can create a truly competitive environment.