The database platform now integrates native AI embeddings, persistent agent memory, faster transaction processing, and private cross-region connectivity to simplify enterprise-scale AI deployment and real-time agent operations.
MongoDB has unveiled a major expansion of its AI-focused database platform at the MongoDB.local London 2026, introducing new capabilities designed to simplify enterprise AI deployment and improve production-scale agent performance.
The announcements strengthen MongoDB’s push toward becoming a unified AI data platform by combining operational databases, vector search, embeddings generation, long-term memory, and data transformation tools into a single stack. The company says enterprises traditionally relied on multiple disconnected systems to run AI agents, creating scalability and integration challenges.
A key launch is Automated Voyage AI Embeddings for MongoDB Vector Search, now in public preview. The feature automatically generates embeddings whenever enterprise data is written or updated, enabling AI agents to retrieve real-time contextual information without requiring separate embedding pipelines. MongoDB said the underlying Voyage AI models currently rank highly on retrieval benchmark tests, improving semantic search accuracy for enterprise applications.
The company also announced general availability of LangGraph.js Long-Term Memory Store integration. The feature gives JavaScript and TypeScript developers persistent cross-session memory for AI agents directly inside MongoDB Atlas, eliminating the need for external memory databases or synchronization infrastructure.
Alongside AI-focused updates, MongoDB introduced performance upgrades in MongoDB 8.3. According to the company, the latest version delivers up to 45% faster reads, 35% faster writes, 15% higher ACID transaction throughput, and 30% better performance for complex operations compared to MongoDB 8.0, without requiring application-level code changes.
The platform expansion also targets regulated industries such as banking, healthcare, and government sectors where deployment flexibility and compliance remain critical. MongoDB announced cross-region connectivity support for AWS PrivateLink, allowing Atlas database traffic between AWS regions to remain entirely on Amazon’s private network instead of the public internet.
Additional releases include Feast Feature Store integration for machine learning workflows, new query expressions for in-database data transformation, and AI skill badge programs aimed at enterprise developers.
MongoDB said the overall objective is to reduce infrastructure complexity for enterprises building production-grade AI systems while improving real-time retrieval speed, reliability, and deployment flexibility across cloud, hybrid, and on-premises environments.
