Sleep-Time Compute
Letting models think during downtime - background agents use idle time to process information and rewrite their memory state between tasks.
Plain English
The agent does its filing at night so the next conversation starts with a tidy desk instead of a replay.
Example
After each support chat ends, a sleep-time agent reorganizes the transcript into durable customer memory - preferences, open issues - so the next session starts curated.
Why it matters
It reframes latency as a schedulable resource: memory consolidation and pre-computing likely next requests move out of the user-waiting window entirely.