Couchbase has expanded its AI Data Plane with native Amazon Bedrock integration and semantic caching, managed Agent Memory and a managed MCP server on AWS, and vector search in Couchbase Operational Insights. All of it is generally available now.
The aim is to lower what production AI agents cost to run. McKinsey has found that moving AI from pilot to production can nearly quadruple costs, and KPMG research in the UK found about a third of business leaders cite limited understanding of AI usage costs as a barrier to deploying agents.
Couchbase said semantic caching through the Bedrock integration has been shown to cut LLM calls by 40% to 60%. Agent Memory, benchmarked against alternatives on the LoCoMo standard, reduces token usage by 33% to 45%, and vector-driven context retrieval produces prompts 70% smaller than full-context approaches. Early adopters report cutting agent deployment times from several months to weeks.
Agent Memory and the enterprise-supported MCP server launched in June as self-managed features. They are now offered as fully managed cloud services on AWS, and the company says both work with LangGraph, CrewAI and LlamaIndex. Vector search in Operational Insights lets teams combine semantic and structured analysis in a single SQL query, with no ETL pipeline, and write the results back to the operational database.
Barry Morris, Chief Product and Strategy Officer at Couchbase, said: “Customers building production AI agents at scale need a converged data layer, but most are still running three or four separate systems for memory, vectors, operational data, and analytics, each carrying its own sync logic, security model, and failure mode.” He said the new capabilities sit on one governed platform that runs the same way in the public cloud, a private data centre or at the edge.
Mavenir, which has run subscriber profiles and live session state on Couchbase since 2017, is extending that deployment to support AI agents for subscriber services and network operations. Brandon Larson, SVP Product Management at Mavenir, said: “Couchbase is a big reason a server failure doesn't drop a subscriber's call.”
To stay across the latest in cloud, AI and enterprise tech analysis from Compare the Cloud, subscribe to our weekly newsletter at https://www.comparethecloud.net/newsletter
