Can AI Agents Ship Code Without Human Review?
AI coding agents are getting better at writing software, but can we trust them to deploy code without human supervision? And what happens when an AI agent makes a mistake in production?
October 2026
In the latest episode of the Singula Team podcast, we talk to Pavel Sviridov, CTO at Nebius Education, about the practical challenges of AI-assisted development, the future of software engineering, and how AI is changing the way technical teams work.
Can We Trust AI-Generated Code?
AI agents can already handle tasks that once required hours of developer time. But giving them more autonomy introduces new risks, especially when production systems and sensitive data are involved.
Drawing on his experience in cybersecurity and software engineering, Pavel shares stories of AI agents making dangerous mistakes and discusses whether AI-generated code is ready to go straight into production.
We also explore how AI is changing cybersecurity, why code quality still matters, and how domain-driven design can help teams create architectures that work better with AI agents.
The Future of Software Engineers
As AI takes over more programming tasks, the role of software engineers is changing. But does that mean companies will need fewer developers?
Pavel shares his perspective on the skills engineers will need to stay relevant, from architectural thinking to understanding complex systems.
The conversation also covers how AI is reshaping technical hiring. Why did Pavel's team remove coding tests from interviews? How should companies evaluate developers who rely on AI agents? And if junior engineers spend less time writing code, how will they gain the experience needed to become tomorrow's seniors?
From Personal AI Tools to a New Way of Working
Beyond software development, AI is changing how people learn, organize information, and build tools for themselves.
Pavel explains how he uses Claude and Obsidian to create a personal knowledge system, or second brain, and how AI agents have become part of his daily workflow as a CTO.
We also discuss the rise of personalized software. As AI makes application development more accessible, could anyone build custom tools instead of relying on traditional off-the-shelf products?
Finally, Pavel reflects on career decisions, leadership, managing complexity, and why building effective systems can matter more than chasing individual goals.
Will AI agents replace developers, or will they redefine what it means to be a software engineer?
Watch the full episode on the Singula Team YouTube channel.
