Context Window

The context window is the AI's working memory: everything it can 'see' while answering — your instructions, pasted documents, and the conversation so far. Modern windows range from a short report to several books' worth of text, and anything outside the window effectively does not exist for the model.

Executives should care because the context window explains two everyday behaviours. First, why pasting the relevant document into the chat improves answers so dramatically — the model reads what is in front of it far better than it recalls training data. Second, why long conversations degrade: as content grows, earlier details can fall out of the window or get lost in the noise, and the AI starts contradicting itself or forgetting instructions.

Concrete example: a team analysing a 300-page tender either pastes the key sections directly, or uses a RAG setup that retrieves the right passages automatically — both are ways of managing the context window deliberately.

Practical rule: give the AI exactly the material it needs for the task at hand, and start a fresh conversation for a fresh task rather than reusing an endless one.