Connect AI documentation · Full API reference & configuration details

Caching for agent responsiveness

Agents often re-issue the same lookups inside a single session. The same account record, the same currency table, the same opportunity stage list. Enable caching on the connection so repeat queries hit a local store instead of the source API, and you'll cut both latency and rate-limit pressure on the upstream system.

Caching is per-connection and configurable per table; pick short TTLs for transactional data (orders, tickets) and long TTLs for reference data (currencies, stage definitions).

References: Caching for concepts and configuration, and Create Cache Connection for the API.