How to Scale Your LLM Usage

A new paradigm of "usage-based scaling" is proposed to increase productivity with Large Language Models (LLMs), building on the successes of scaling in pre-training, post-training, and inference. This approach encourages professionals, especially programmers, to maximize their use of AI by running multiple agents and tasks concurrently. Specific techniques suggested include running parallel coding agents for bug fixes or feature development, initiating deep research tasks on topics of interest, and creating automated workflows with tools like n8n. The core principle is to leverage the decreased effort required to start tasks by continuously running useful AI processes in the background.