Term: Context window
~6 min read
Estimated time: ~6 min read — for the in-app brief plus opening the primary source.
What this is
The context window is how much text the model can consider at once, measured in tokens — your prompt, attachments, chat history, and its reply share the same budget.
Everyday example
Paste a 120-page contract into a chat tool and it may “forget” early clauses or refuse the upload. The context window is full — a capacity limit, not a personal slight.
The context window is how many tokens fit in one sitting — prompt, files, history, and the reply share it.
- When the window fills, early content is ignored or the upload fails.
- A long chat slowly “forgets” the start. That is capacity, not mood.
- Retrieval (RAG) exists because whole-file paste does not scale.
- Windows differ by product and plan — check yours.
Next action: For any “paste the whole contract” habit, switch to section-by-section or retrieval.
What changes in how you lead
How decision rights, process, and ownership should change.
- Do not promise a workflow that assumes infinite memory.
- Operations designs around the window, not around the demo.
Compare related ideas
Context window vs Token
Tokens are the units. The context window is the bucket they fill. Filling the bucket is why long chats degrade or uploads fail.
Open TokenContext window vs RAG
A bigger window lets you paste more at once. RAG fetches the relevant pieces so you do not rely on stuffing everything into one prompt.
Open RAGDeep dive
Larger windows help long documents. They still are not infinite memory of your company, and they cost more to run.
Legal and Finance: due diligence and board packs usually need retrieval or sectioning, not one paste.
Marketing: keep the live brief short; pull brand rules from an approved store (RAG).
Operations: send the last N incident events, not the full year.
ChatGPT, Claude, Gemini, and Grok advertise different window sizes. Product limits change — verify for your use case.