Please use the form below to provide your feedback. Because your feedback is valuable to us, the information you submit in this form is recorded in our issue tracking system (JIRA), which is publicly available. You can track the status of your feedback using the ticket number displayed in the dialog once you submit the form.

Agent Memory Release Notes

  • reference
New features and changes in Couchbase Agent Memory for Managed and Self-Managed Agent Memory.

For changes to the Agent Memory API and the Capella Management API, see the API change logs for Managed Agent Memory and Self-Managed Agent Memory.

October 2026

Managed Agent Memory

Capella now offers a fully managed Agent Memory server, so AI agents can store and recall what users told them across sessions. Agent Memory extracts facts, embeds them, and makes them semantically searchable in your Capella operational cluster, with no memory infrastructure for you to host or maintain.

With Managed Agent Memory, you can:

Managed Agent Memory is available on AWS in the US East (N. Virginia) and Europe (London) regions.

Managed Agent Memory runs Agent Memory 1.1.0 and includes all of its features. For more information, see Agent Memory 1.1.0.

Agent Memory 1.1.0

Agent Memory 1.1.0 is available for Self-Managed Agent Memory and adds the following features:

Multi-node scaling

Agent Memory can now run on multiple nodes, with automatic recovery if a node fails. For more information, see Cluster Mode.

Annotation-based filtering

Agents can scope recall to a tenant, source, or channel. For more information, see Use Annotations.

Custom extraction prompts

Register your own extraction prompts to steer what each agent remembers. For more information, see Customize Extraction and Summarization Prompts.

Separate administrator and application roles

Give your agents only the access they need to store and search memory, and reserve administrative operations, such as managing custom prompts, for administrators. On Self-Managed Agent Memory, you assign roles through your identity provider. On Managed Agent Memory, you choose an access level for each API key. For more information, see Configure RBAC for Self-Managed Agent Memory and Control Access to Managed Agent Memory.

Memory provider for Hermes Agent

You can now use Agent Memory as a memory provider for Hermes Agent. For more information, see Agent Memory Plug-in for Hermes Agent.

June 2026

Agent Memory 1.0

Self-Managed Agent Memory is now available. Agent Memory gives your agents a persistent memory layer that stores conversation history, extracted facts, and vector embeddings in a Couchbase cluster.

For more information, see Agent Memory for Persistent Memory Storage.