红杉资本：这就是AGI

Core Insights - The emergence of "Long-horizon agents" marks the arrival of Artificial General Intelligence (AGI), with 2026 being identified as a pivotal year for its maturity [1][2] - The transition from software sales to selling work outcomes signifies a fundamental shift in business paradigms, as AI evolves from being a mere tool to a digital employee capable of autonomous work [2][7] Functional Definition - AGI is pragmatically defined as the ability to "self-solve problems," focusing on the AI's capability to complete tasks rather than the technical definitions surrounding it [3] - Three core elements of this capability include baseline knowledge (pre-training), reasoning ability (computational reasoning), and iterative ability (Long-horizon agents) [3] Autonomous Work Loop - Long-horizon agents can autonomously execute complex tasks, exemplified by their ability to conduct comprehensive recruitment processes without human intervention [4][5] - These agents can analyze various data sources and adapt their strategies to achieve recruitment goals, showcasing their problem-solving capabilities in ambiguous environments [5] Technological Pathways - The advancement of AGI is driven by two main technological pathways: Reinforcement Learning and Agent Harnesses, both of which have shown scalability and effectiveness [6] - Current trends suggest that by 2028, intelligent agents will reliably complete tasks that currently require a full day of human work, with projections extending to completing a year's worth of work by 2034 [6] Business Transformation - The rise of specialized intelligent agents across various sectors indicates a significant paradigm shift for entrepreneurs, moving from dialogue-based AI applications to execution-focused agents [7] - This transformation necessitates a reevaluation of how tasks are managed and priced, emphasizing outcomes rather than tools, as the market prepares for the exponential growth of long-horizon agents [7]