From PR to AI review: building a multi-agent code review system

Modern AI tools are accelerating software development, but most teams still rely on fragmented workflows and manual code reviews.

In this session, Nikolay Advolodkin will walk through how he built a multi-agent AI system that automatically reviews pull requests, combining tools like Claude Code, GitHub Copilot, and CI pipelines into a structured engineering workflow.
Instead of focusing on individual tools, this session shows how to design a system where multiple AI agents analyze code from different perspectives, aggregate results, and produce actionable feedback directly in pull requests.
You’ll see how the system works end-to-end, how the tools fit together, and what actually worked (and didn’t) when building it.

Nikolay Advolodkin is the CEO and Chief Engineer of UltimateQA, where he has trained over 150,000 engineers across 190 countries. He specializes in building distributed, resilient agentic systems that leverage AI to accelerate software development and testing.
His work has delivered measurable results for Fortune 500 companies, including up to 97% reductions in execution time and 560% performance improvements.
He is the creator of NoteMyLife.com and a global speaker on AI-driven development, focused on helping engineers design systems that integrate tools like Copilot and Claude into real production workflows.
His mission is to elevate human lives through technology by replacing friction with leverage.