AI Leader Blog

Yesterday Anthropic yielded to common conventions over exclusive proprietary instructions, and adopted AGENTS.md as a standard location for placing instructions targeted at AI agents. In general, open, shared conventions are helpful. The larger problem is that the idea of AGENTS.md indicates a deeper issue: the lack of disciplined documentation whether directed at humans or AI assistants. Trying to cordon off the problem instead of addressing it head-on is short-sighted, and will only perpetuate the “knowledge gap” that most projects carry already.

Much has been made of the ability of LLMs to write simple applications from scratch. Less has been written about how well LLMs can maintain an existing project. Working with an existing codebase exposes something that is both a strength and a weakness of an LLM: following patterns. The least wrinkle in a codebase can cause an LLM to err through pattern-following, overriding both its training and its reasoning. An existing pattern can throw a powerful model such as Claude Opus 4.8 off track even on something as simple as a function to parse an integer.

At the most fundamental level, Large Language Models (LLMs) do not actively reason — they predict. LLMs use probability to produce what someone would have likely said in response to a prompt. They do not consider whether the response is correct as they are writing it. However the same statistical output means they can actually do a good job of reviewing their own output, based upon what someone probably would say having reviewed that work.