What it is
A context window is the finite span of tokens (prompt, history, retrieved data) an LLM can hold in active memory at once. Once exceeded, older content gets truncated or summarized, which can cause an agent to 'forget' earlier instructions.
Why it matters
Context window size directly limits how much grounding data, conversation history, and instructions an Agentforce agent can use before quality degrades. Choosing the right model and context strategy affects cost, accuracy, and reliability.
Key components
- token budget
- sliding window
- context truncation
- memory summarization
How it connects
Agentforce and Data Cloud manage context windows by chunking retrieved knowledge via RAG rather than dumping entire documents into a single prompt.
Good to know
Bigger isn't always better — larger context windows increase latency and cost, so smart retrieval often beats brute-force size.
Related terms
Agent Memory
The systems that let an AI agent retain context across calls, sessions, users, or teams — turning a stateless model into something with continuity. Encompasses short-term working memory, long-term episodic memory, and shared organizational memory.
LLM (Large Language Model)
The AI technology behind ChatGPT, Claude, and the intelligence in Agentforce. Trained on massive amounts of text to understand and generate human language.
RAG (Retrieval-Augmented Generation)
A technique that makes AI smarter by fetching relevant information from your data before generating a response. The AI "looks it up" instead of guessing.
Token
The basic unit of text that AI models process — roughly 3/4 of a word. Token limits determine how much context an agent can use in a single interaction.
