Factory’s Matan Grinberg on Recursive Self-Improvement, Chinese Models & Lessons From String Theory

Autonomous software engineering is fundamentally reshaping development workflows as AI agents transition from simple coding assistants to independent problem-solvers. Matan Grinberg, founder of Factory, explains that the shift toward agentic systems allows enterprises to automate complex tasks, such as legacy COBOL migrations, in days rather than years. The primary bottleneck in modern engineering is no longer code generation but rather the strategic decision-making required to deploy these agents effectively. Furthermore, the discussion addresses the critical need for balanced AI regulation, advocating for open-source models to prevent monopolistic control by closed-provider labs while acknowledging the security risks inherent in long-term AI integration. Ultimately, the transition to an AGI-enabled economy demands a shift in hiring priorities, favoring high-agency individuals capable of managing autonomous systems over traditional algorithmic coding skills.