Glossary
Context Window
The maximum amount of text a model can consider at once, including your prompt, any documents you provide, and its own reply.
The context window is the model's working memory for a single request. Everything it should take into account, your instructions, the conversation so far, and any reference material, has to fit inside it, measured in tokens.
Bigger windows let you hand over more at once, but they aren't a substitute for good retrieval. Stuffing in everything raises cost and can bury the important details, so feeding the model the right context usually beats feeding it the most.
How we use it
We design what goes into the context window deliberately, using retrieval to supply just the relevant material, which keeps answers sharp and costs predictable.
Get in touch
Want to put this into practice?
If this concept is relevant to something you're building, a short note is the fastest way to get practical help.