Couchbase expands AI Data Plane with caching and agent memory

Couchbase announced native Amazon Bedrock integration with semantic caching, a managed Agent Memory and MCP server on AWS, and vector search in its Operational Insights engine. The company says caching cuts LLM calls by 40–60% and Agent Memory reduces token usage by 33–45%.

• Semantic caching cuts LLM calls by 40–60%, Couchbase says
• Agent Memory reduces token usage by 33–45% on LoCoMo
• Vector search in Operational Insights yields prompts 70% smaller
• Agents generate 450% more traffic per task than humans