Sber presents "GigaAgent": an autonomous AI assistant that writes its own code
The Russian AI agent market is entering a new phase. Following announcements from Yandex, Sber has officially launched its own universal autonomous assistant — "GigaAgent." This is not just another chatbot, but a digital partner capable of independently planning and executing multi-step tasks, marking a fundamental shift from conversational interfaces to proactive systems.
From dialogue to action: three key differences
GigaAgent can manage a calendar, work with documents, handle correspondence, search for information, and prepare reports with presentations. However, the main thing is its architectural principles. Unlike traditional models, this agent possesses three critical properties: autonomy, self-development, and continuity. It operates in the background without requiring control at every step, and can even reasonably object to the user when data is insufficient.
The most intriguing aspect is the self-development mechanism. GigaAgent can analyze and rewrite its own source code, recording each change as commits in a built-in repository. Additionally, it preserves context between sessions, which distinguishes it from most AI systems that operate within a single session and forget accumulated experience.
Technological foundation and cost
At its core is the Ouroboros project, developed by researchers at the AIRI Institute. On the Terminal Bench 2.1 benchmark for Anthropic's Opus 5 model, its performance quality reaches an impressive 86.97%. The launch is carried out through the Agents Space platform from Cloud.ru, where new users receive a starting grant of 4,000 rubles. Further pricing depends on consumption volume and language model tariffs.
According to expert estimates, development cost 1–3 billion rubles over a year and a half, with the project team comprising about 100 specialists. At the same time, Sber does not disclose exact figures but emphasizes that the base GigaChat 3 model was already released under a free license, which reduced training costs from scratch.
My view: Claims of "self-development" are a powerful marketing move, but the key question lies in measurable business metrics. Whether the agent can actually reduce operational costs and increase productivity will only be shown by practice. For now, we are watching Russian giants lay the foundation for a new era of digital partners, and the one who first proves economic efficiency will gain a decisive advantage.