Personal superintelligence

Garry Tan argues that personal AGI comes from combining a rented frontier model with a person’s own context and an agent harness. He explains how to build a personal knowledge library and reusable skills, using agents to expand individual capacity and automate knowledge work. He emphasizes keeping that memory and those skills under personal control, with careful curation and security. The talk closes with practical steps for starting small and a case for making powerful tools broadly available.

Alexandr Wang recounts how early work at Quora and MIT led him to found Scale around the then-overlooked need for training data. He argues that AI progress is already sufficient to create decades of economic change, and that builders can use agents to help startups compete with much larger companies. At Meta, he describes a vision of personal superintelligence, a rapidly rebuilt frontier lab, and more accessible models and tools for developers. His advice is to develop a strong view of the future, think systematically as the abstraction layer shifts toward coordinating agents, and invest in steep, long-running technological trends.