Descriptions:
Matan Grinberg, CEO and co-founder of Factory (an AI-native software development platform) and former string theorist, joins Harry Stebbings on 20VC to discuss how AI is fundamentally reshaping engineering organizations, team structures, and the economics of software development. Grinberg argues that meaningful productivity gains from AI tools are inevitable but slow to permeate organizationally — companies must first decide whether to solve more problems or solve the same problems more efficiently, a question he believes most enterprises are still working through.
The conversation digs into the future of engineering talent, engaging with Andrej Karpathy’s “100x engineer” thesis. Grinberg reframes it around what he calls “load-bearing individuals” — people whose removal causes things to fall apart — arguing these high-leverage contributors now wield even greater impact with AI tools, widening the gap between top and average performers. He also addresses the frontier model debate, suggesting enterprises may see a short-term contraction in usage of the very best models as cheaper alternatives prove sufficient for routine work.
Grinberg covers the evolution of AI code review — from the first wave of “generate everything” enthusiasm to the “phase two” problem of staff engineers buried under poorly formatted AI-generated PRs — and explains why investing in CI/CD pipelines, linters, and pre-commit hooks pays off disproportionately in an agent-native development environment. He also discusses how Factory’s customers are using agents to automate previously manual tasks like release notes and documentation, work he expects to look embarrassingly archaic within five years.
📺 Source: 20VC with Harry Stebbings · Published June 13, 2026
🏷️ Format: Interview
