Sber presents "GigaAgent": a new round in the race of autonomous AI agents in Russia
The Russian artificial intelligence market is entering a new phase. Sber has officially announced the launch of "GigaAgent" — a universal autonomous assistant designed to handle both everyday and complex work tasks. This is a direct response to intensifying competition in the AI agent segment, where Yandex previously introduced a similar product. We are witnessing a shift from simple conversational chatbots to full-fledged digital partners capable of planning and executing multi-step processes.
A breakthrough in autonomy and self-learning
"GigaAgent" stands out from existing solutions due to three key characteristics. First, full autonomy: the agent operates around the clock, in a proactive mode, can request clarifications when data is insufficient, and can reasonably challenge inputs it deems incorrect. Second, self-development: the system analyzes and modifies its own source code, recording each change as a commit in a built-in repository. This allows it to improve without direct developer intervention.
Third, continuity. Unlike most AI agents that operate within a single session and lose context, "GigaAgent" retains memory of past tasks, decisions made, and interaction history. After a restart, it continues working with accumulated experience, which dramatically increases the efficiency of long-term projects.
Technological foundation and cost
The development is based on the Ouroboros project, created by researchers at the AIRI Institute. On the Terminal Bench 2.1 benchmark, Anthropic's Opus 5 model with this agent achieves an impressive score of 86.97%. The service is launched via the Agents Space platform from Cloud.ru. New users receive a starting grant of 4,000 rubles to test the capabilities, after which usage is billed depending on the language models and infrastructure used.
Estimates of investment in the project vary: according to experts, orchestration, connectors, interface, and the quality assessment system required from 1 to 3 billion rubles and a team of about 100 specialists. For Sber, this is a relatively small amount, especially given that the base model GigaChat 3 was previously released under a free license, eliminating the costs of training from scratch.
My analysis
The launch of "GigaAgent" is not just a technological release, but a strategic bet on leadership in the fundamental transformation of the labor market. However, the key challenge remains the measurability of effectiveness: behind the loud claims of self-development, there must be clear business metrics and understandable control mechanisms; otherwise, autonomy could become a source of unpredictable risks. For the Russian market, this is an important step toward creating a mature AI ecosystem capable of competing with foreign counterparts.