Yandex has announced the launch of a large-scale "75/75/75" program, aimed at integrating artificial intelligence into development standards at a fundamental level by the end of 2026.
What Happened
According to the new strategy, 75% of the company's developers should regularly use AI, 75% of all code changes should involve neural networks, and 75% of the code in such changes should be AI-generated. To achieve these goals, specialized tools are already in use: the GENA autonomous agent for writing code and the Stefania virtual employee for analytical tasks, which is currently used by more than 5,000 employees.
Context
Instead of using neural networks as isolated auxiliary tools, the company is moving toward an AI-native development model. This implies transforming AI from a simple autocomplete tool into a full participant in the software development lifecycle, where automation covers not only writing lines of code but also analytical processes.
Why It Matters for the Industry
Yandex is setting aggressive quantitative metrics that could become a new industrial benchmark for large technology companies. Demonstrating working patterns for implementing AI agents in enterprise structures shows the potential for real and measurable reductions in Time-to-Market for critical components.
Why It Matters for Users
For engineers and developers, this signifies a paradigm shift: the human role is shifting from direct code writing to architectural control, managing AI agents, and validating their results. This is a transformation into a "Reviewer-centric workflow," where the key competency becomes the quality control of generation.
What Is Not Yet Known / Limitations
There is a critical risk associated with the program's success depending on the effectiveness of quality control mechanisms and process management once the target level of 75% code generation is reached.
Sources
Author
Look at AI, Editorial Team