How our site is built for LLMs

When we set out to launch Theory’s new site, we quickly realized that in the age of large language models, our most prolific “readers” are no longer people but software—chatbots, copilots, and indexers -  that distill and retrieve our content into bite-sized answers for founders, LPs, and reporters. 

We'll be maintaining llms.txt and llms-full.txt files for agent-based browsing. Similarly, each blog post will exist as a .md file for agent-memory augmentation or RAG systems. 

Finally, we refused to sacrifice the human layer. The design system uses generous light space, high-contrast typography, and motion only where it adds comprehension. 

The result is a site that welcomes two very different audiences in tandem: humans who want inspiration and context, and machines that crave atomic facts. We hope these choices will serve users via a browser or LLM interface.