Sequoia Ascent 2026 Summary (karpathy.bearblog.dev)

At the Sequoia Ascent 2026 event, a significant discussion revolved around the evolution of AI agents and their transformative impact on programming and software development. The speaker emphasized that we have entered an era termed "Software 3.0," where large language models (LLMs) are not just tools for generating code but are becoming integral as a new programmable layer for digital work. This shift has made programming a practice of orchestrating LLMs to perform complex tasks rather than solely writing lines of code. The discussion highlighted a pivotal moment in late 2025 when these agents began producing larger, more coherent outputs, fundamentally changing the workflow. Key concepts introduced include "verifiability" in automating tasks and the distinction between "vibe coding," which encourages accessibility in software creation, and "agentic engineering," which requires a deeper understanding of how to manage and leverage these AI tools. The speaker called for a reevaluation of hiring practices to emphasize skills relevant to managing AI agents effectively, asserting that opportunities lie in uncovering domains ripe for AI integration that are currently under-exploited. With the focus shifting towards agent-native infrastructures and understanding the distinct nature of LLMs, the message is clear: the AI landscape is rapidly evolving, and adapting to these changes will redefine programming roles and educational approaches in the tech industry.