Smart Memory

Injecting every memory into every prompt would waste tokens and dilute the answer. Smart Memory selects only the memories relevant to the current conversation.

Why it matters#

Lean context means lower token cost and sharper answers - your AI gets what it needs for this conversation, not your entire memory. This reduces token injection substantially compared with dumping all context into every prompt.

How it fits#

Smart Memory is the recall half of the model described in how it works. It draws from the right bucket and injects only what is relevant.

Related#

• Buckets
• How it works
• Back to docs home