Context window
The context window is the amount of text an AI model can hold in mind at once, your prompt, the conversation so far, any documents you've pasted, and its own reply, all counted together in tokens. Once you exceed it, the oldest content falls out of view and the model effectively forgets it. Models range from a few thousand tokens to over a million.
For a founder this is the practical limit on how much an agent can juggle in a single task. Feed it a whole knowledge base and it may run out of room or, worse, lose the important detail in the noise, models attend less reliably to the middle of a long window. It matters for cost too: every token in the window is billed on every call, so a bloated context gets expensive fast. The skill is feeding the model exactly what it needs for the task, not everything you have.