Karpathy, Google, Tan agree Markdown is the answer, but not for the same problem (thenewstack.io)

Andrej Karpathy, Google, and Garry Tan have all converged on Markdown as a vital resource for AI agents, each tackling different challenges in the AI/ML ecosystem. Karpathy initiated this trend in April with his "LLM Wiki," proposing that AI agents utilize linked Markdown files for personal knowledge management. Two months later, Google responded by introducing the Open Knowledge Format (OKF), which standardizes organizational knowledge in Markdown for use in AI applications. Meanwhile, Tan's gstack has garnered significant attention, serving as a collection of specialized Markdown files that streamline interactions with multi-agent systems. The significance of this development lies in the shifting paradigm of competitive advantage in AI. Rather than the quality of models being the sole determining factor, the emphasis is now on the Markdown documentation that teams create and maintain over time. This transition from model dependency to the ownership of easily portable Markdown files suggests that documentation could serve as a long-term advantage in AI deployments. However, Google's OKF is still in its nascent stages, highlighting the need for robust consumer development to transform this idea into a sustainable standard. If effectively adopted, Markdown could become the foundational format through which AI agents interpret context and deliver insights across various platforms and models.