MongoDB has announced a set of new capabilities designed to move enterprise AI from experimental deployment into reliable production, addressing what the company describes as the critical but often overlooked data infrastructure layer beneath AI agents.

The announcements were made at MongoDB .local London 2026 and are directly relevant to Singapore enterprises advancing from AI pilots to organisation-wide deployments.

What’s New: Embeddings, Memory, and Performance

The centrepiece of the release is Automated Voyage AI Embeddings in MongoDB Vector Search, now in public preview. The capability generates embeddings automatically as data is written or updated, removing the need for manual infrastructure configuration and enabling semantic search to be deployed in minutes rather than weeks.

MongoDB’s Voyage AI embedding models currently rank first on the Retrieval Embedding Benchmark (RTEB), a key measure of how accurately an AI agent can surface relevant information from a data store.

For JavaScript and TypeScript developers, the LangGraph.js Long-Term Memory Store integration — now generally available — provides persistent, cross-conversation agent memory backed by MongoDB Atlas, without requiring an additional database. The capability brings parity with Python developers, who have had access to similar functionality.

MongoDB 8.3 Delivers Significant Performance Gains

MongoDB 8.3, available immediately, delivers up to 45% more reads, 35% more writes, 15% more ACID transactions, and 30% more complex operations compared with MongoDB 8.0 — without requiring changes to application code. The update also moves common data transformations into the database itself, removing the need for external pipelines to feed AI agents.

Cross-region connectivity support for AWS PrivateLink is also now generally available, keeping database traffic between MongoDB Atlas clusters on the AWS private network and helping security and compliance teams approve cross-region architectures with fewer exceptions.

One Platform Across Cloud, On-Premises, and Hybrid

MongoDB runs across Amazon Web Services, Google Cloud, Microsoft Azure, on-premises, and hybrid environments — a key consideration for regulated industries in Singapore and across APAC where data residency requirements often constrain deployment choices.

“The hardest part of running agents in production isn’t the model. It’s the data layer underneath it,” said CJ Desai, President and Chief Executive Officer of MongoDB. “To trust an agent at scale, it has to retrieve the right context, hold memory across sessions, and operate at machine speed, wherever the enterprise needs it.”