XNews

Databricks Launches Agent Memory on Agent Bricks with Lakebase Access

Databricks Launches Agent Memory on Agent Bricks with Lakebase Access

Databricks announced Agent Memory on Agent Bricks, combining session history with distilled long-term memories. A background pipeline extracts durable facts, removes duplicates, and replaces contradicted information, with Lakebase providing low-latency access. Internal benchmarks of enterprise data tasks showed the architecture tripled agent pass rates while using 31% fewer tokens than agents without memory.

Event version 1.

Key insights

Memory architecture combines session and long-term data

Agent Memory on Agent Bricks combines session history with distilled long-term memories. A background pipeline extracts durable facts, removes duplicates, and replaces information contradicted by newer conversations.

Lakebase provides low-latency access

Lakebase provides low-latency access for the Agent Memory architecture.

Benchmarks show tripled pass rates and token savings

In internal benchmarks of enterprise data tasks, the memory architecture tripled agent pass rates while using 31% fewer tokens compared to agents without memory.

Source posts