New on the IBM Db2 Community: Persistent AI Agent Memory with Db2 12.1.5 VECTOR

Jason Perlow

Jason Perlow

— 1 min read
Share
New on the IBM Db2 Community: Persistent AI Agent Memory with Db2 12.1.5 VECTOR
Cloud “memory” features exist, but they store your context somewhere outside your infrastructure. I wanted something different.

I recently published a new hands-on article for the IBM Db2 Community that demonstrates how to build persistent AI agent memory using IBM Db2 12.1.5 VECTOR and my open-source project, mnemos.  

The guide walks through deploying a complete local stack with Db2 Community Edition, configuring native vector search with VECTOR, VECTOR_DISTANCE, and DiskANN indexes, and connecting AI assistants such as Claude Code, Cursor, OpenAI Codex, Claude Desktop, and ChatGPT through the Model Context Protocol (MCP).

Rather than relying on a separate vector database or cloud-hosted memory service, mnemos stores agent memory directly in the enterprise database your organization already operates, providing durable, auditable, semantic recall across AI tools.

If you’re interested in enterprise AI, vector search, or building coding assistants that remember what happened in previous sessions, take a look.

Read the full article on the IBM Db2 Community: Persistent AI Agent Memory on Db2 12.1.5 VECTOR — mnemos Quick Start

Project: https://github.com/ncz-os/mnemos

Read more