Cian aims to transition its entire software development process to AI agent models by the end of 2026. Currently, neural networks contribute to 75% of the development effort, with over 35% of daily code lines generated with their assistance.
- AI influences 75% of development now, targeting 100% integration by 2026.
- More than 35% of daily code lines are already produced by neural networks.
- AI proficiency is now evaluated in all Product & Technology job interviews.
How AI reshapes the software engineer’s role
At the Data Fusion conference, Cian’s CTO Maxim Radyukov explained the shift: ”Using AI is a qualitative leap that doesn’t replace people but expands their capabilities. Code becomes outdated the moment it hits production. The real value isn’t in lines of code but in framing the right context and validating outcomes. As the industry evolves, code’s value declines, so we’re changing hiring practices – all Product & Technology interviews now test AI skills.”
Cian sees AI as a tool that frees developers from routine tasks, enabling them to focus more on architectural decisions. What used to require cutting corners can now be approached with more thorough and refined solutions.
AI agents perform like mid-level developers with human oversight
The company stresses that with proper training and setup, AI functions at the level of a competent mid-level developer. However, responsibility for the AI’s output remains with the human engineer, who decides which tasks to delegate and which require direct control.
Systematic analysis skills become essential: engineers must define clear requirements, structure them for large language models, and set acceptance criteria.
Quality assurance and security increase with AI automation
As AI-driven development expands, Cian is strengthening quality assurance and information security. Their AI use cases are divided by data sensitivity: local models handle confidential and regulatory tasks, while external AI services manage less sensitive workloads.
Cian’s aggressive AI strategy highlights a broader trend in software development: moving beyond AI as just a tool to full integration as a coding partner. While companies like Microsoft and Google embed AI assistants to boost coding efficiency, Cian aims to push that boundary by automating nearly all code production within a few years.
The shift raises questions about future software engineering careers: Will human developers become primarily architects and validators? Can AI agents handle creative problem solving at scale? Cian’s experiment will be a key case study in whether full AI integration delivers real productivity gains without compromising quality or security.