Self Coding Agents - AI That Builds and Improves Itself

Explore the cutting-edge concept of self-improving AI coding agents in this 17-minute conference talk from the AI Engineer Summit 2025. Discover how Augment Code developed an AI agent that not only writes code but continuously enhances its own codebase, including performance optimizations and system architecture improvements. Learn about the practical challenges and opportunities of supervising AI coding agents at scale, understand the current limitations and scope boundaries of AI coding capabilities, and examine the potential future where human engineers orchestrate multiple self-improving AI agents working in parallel. Gain insights into the framework, testing methodologies, and real-world examples of autonomous AI development, while exploring the implications for the future of software engineering where humans may shift from writing code to driving innovation through AI orchestration.