MemCurrent aims to give production AI agents persistent memory

Databricks introduced MemCurrent to help agents carry user preferences, corrections and other context between sessions. It uses plugins to log and load context, tools agents can call on demand, and an offline pipeline to consolidate memory. Databricks says that on an internal benchmark built from real enterprise data, MemCurrent tripled the agent's pass rate while using 31% fewer tokens and running 26% faster. It is built into Agent Bricks' Memory offering and powers Genie, the company says.