Former GitHub CEO Debuts AI-Native Git Platform With Persistent Memory for Coding Agents

Nat Friedman, the former CEO of GitHub, has unveiled a new Git-based platform designed specifically for AI agents. The system integrates automated code reviews, distributed repository hosting, and a persistent memory layer that allows AI coding assistants to retain context across multiple sessions. This addresses a major limitation of current AI tools, which often lose track of project history and developer intent between interactions. By enabling agents to remember past decisions and code patterns, the platform aims to reduce repetitive prompting and improve code quality over time. The move signals a shift toward infrastructure built for autonomous software development, where AI agents collaborate like human developers. As AI coding tools gain traction, such specialized platforms could become essential for scaling agent-driven workflows. Friedman's venture is likely to attract attention from enterprises looking to deploy AI developers securely and efficiently. The platform's distributed hosting also suggests a focus on resilience and decentralization, key concerns for large-scale AI operations.