Sber has named the next stage of AI development for large businesses: the transition to autonomous agents. Such systems operate independently, completing tasks from start to finish without human intervention, the company announced at the GigaConf-2026 conference, TASS reports.

What does this mean in practice? Previously, AI assisted at individual stages: writing code, finding errors, or generating text. An autonomous agent takes over the entire process. It plans steps, uses the necessary tools, checks results, and delivers a finished outcome. The company believes that large organizations will implement this gradually — function by function, rather than the entire structure at once.

For agent development, Sber has created three families of solutions: GigaAgent, GigaCowork, and GigaCode. A unified repository of skills for these products has been placed on the GitVerse platform. Andrey Belevtsev, Senior Vice President of Sber, noted: a year ago, the quality of such systems did not allow for their mass application, but now generative AI has become an everyday tool for all company engineers.

Many companies have already passed the stage of automating individual operations. The next step is holistic autonomous organizations where agents work without constant supervision. But for large businesses, the transition will be smooth. Understanding what 'autonomous' means here is critical: it's not just a smart assistant, but a digital employee to whom a task can be fully delegated. If such systems meet expectations, they will change the very approach to product development and process management.