Surprisingly Useful AI Article Enhancements
MongoDB has unveiled a set of capabilities designed to give AI agents direct access to real-time operational data, addressing what the company describes as persistent complexity around embedding pipelines and data synchronisation.
The announcements, made at MongoDB.local Build Fest in San Francisco, centre on two areas: automated embedding within the Atlas platform powered by Voyage AI models, and a fully hosted MCP (Model Context Protocol) server that connects coding agents to Atlas without requiring teams to run additional infrastructure.
Automated Embeddings Remove Pipeline Overhead
Until now, developers building retrieval-heavy AI applications have typically had to extract data from operational systems, manage embedding pipelines and keep vector stores synchronised with live databases. MongoDB argues this approach is fragile and introduces latency.
With Automated Embeddings in MongoDB Atlas, powered by Voyage AI, developers can insert and retrieve data while Atlas handles embedding and indexing in the background. When new documents are written or existing ones are updated, the system re-embeds them automatically without manual scheduling or external tooling.
MongoDB’s Voyage AI embedding models currently hold first place on the Retrieval Embedding Benchmark (RTEB).
The release also includes voyage-code-4, a retrieval model designed specifically for coding agents. According to MongoDB, general-purpose embedding models were not built for codebases and underperform on code retrieval tasks. The new model is tuned to match information to code with higher precision and lower cost than earlier models.
A standalone Atlas Embedding and Reranking API gives applications running outside MongoDB access to the same Voyage AI embedding and reranking models through Atlas.
Vector search in Atlas Stream Processing extends retrieval accuracy to data in motion, so agents acting on live event streams can work with the same context quality as those querying data at rest.
Financial Times And Eve Among Early Users
The Financial Times is among the organisations already using the automated embedding capability. The newspaper had seen its search functions splinter across multiple teams and products, and consolidated onto the MongoDB platform to address the issue.
Principal Engineer CM Platform at the Financial Times, Elitsa Pavlova, explained: “Our job is to make the FT’s journalism fast and easy to reach, however our readers come to it. With Automated Embedding and Voyage AI models on Atlas, we’ve improved retrieval accuracy while keeping costs in check across more than 100,000 searches a day, and being able to test and balance models lets us tune quality against cost as we go.”
“With less infrastructure to run, the team can spend more of its time on the reading experience our subscribers rely on,” Pavlova added.
Legal AI platform, Eve, is using the Atlas Embedding and Reranking API to surface relevant material across the lifecycle of legal cases.
Staff AI Engineer at Eve, Urvesh Patel, commented: “In legal AI, retrieval quality is foundational – the right evidence has to surface at the right moment. MongoDB’s Atlas Embedding and Reranking API gives us a promising way to improve relevance directly in the RAG layer, while simplifying the infrastructure needed to build and evolve these experiences.”
MongoDB Positions Atlas As An Agent Memory Layer
MongoDB frames Atlas as a combined memory and context layer for AI agents, with retrieval running against the same live operational data rather than a separate, synchronised copy.
Chief Technology Officer at MongoDB, Jim Scharf, pointed to the complexity many organisations face when running AI workloads in production.
Related: Best Business Laptops for work & school
Related: Best Gaming Laptops
Related: Best Portable Laptop
“Too many organisations are running AI in production with an operational database, a vector store, a search engine, and embedding and reranking models, all from different vendors, bolted together instead of built for it,” Scharf noted. “That’s where stale data and errors creep in, and it’s usually where teams spend their time babysitting instead of building.”
“Agents raise the bar. They need to retrieve live context continuously and cannot wait on overnight batch jobs. MongoDB was built as an operational platform from the start, so retrieval and memory run on the same live data, nothing to sync, and agents act on what’s happening instantly,” he added.
Managed MCP Server Connects Coding Agents To Atlas
The second major component of the release is the MongoDB Atlas Managed MCP Server, a fully hosted service that connects AI agents to Atlas data without requiring teams to install or maintain local infrastructure.
MongoDB’s existing MCP server already records more than 30,000 installs a week. The managed version runs as a service inside Atlas, using the same credentials and access controls that teams already have in place. Administrators can govern how agents access operational data from a single point.
Once connected, agents can list collections and indexes, query and aggregate data and inspect schemas. With appropriate permissions, they can also create collections or manage indexes.
Chief Product Officer for AI and Emerging Products at MongoDB, Pablo Stern-Plaza, remarked: “The AI tools teams reach for keep changing, so our approach is to make sure MongoDB is present in all of them, whether a team is working in Claude or Codex, or running an agent in production.”
“More of that building is now done by agents, and neither the agent nor the developer has to stop and set up a connection, so applications come together faster,” Stern-Plaza continued.
Integrations Span Claude Code, Codex, Grok Build And Devin
MongoDB has also introduced native integrations across a range of AI development tools. Developers can now connect Atlas to Claude Code, Codex, Grok Build and Devin through marketplace connectors, with no connection strings to paste or infrastructure to configure.
The integrations allow users to query data in plain language through ChatGPT, Claude and Grok, while coding agents such as Claude Code, Codex, Grok Build and Devin can query, inspect and update MongoDB data as part of their workflows. Developers using IDEs like Cursor can also view live MongoDB data while generating applications.
Product Lead for the ChatGPT Ecosystem at OpenAI, Vibhor Chhabra, observed: “Developers want their AI tools to connect with the context and systems they already rely on. MongoDB’s plugin in ChatGPT makes it easier to access and work with live application data, helping developers move faster while staying grounded in the context of their applications.”
President at Cognition, Russell Kaplan, whose company builds the Devin coding agent, offered a similar perspective: “The scope of what one engineer can build has exploded, and the unlock is agents working with real context. By connecting Devin to MongoDB Atlas, engineers can hand off well-scoped tasks knowing Devin is working from live application data, not stale assumptions, and spend their own time on the harder problems.”
Availability
The new capabilities, including Automated Embeddings on Atlas, the Atlas Embedding and Reranking API, voyage-code-4, vector search in Atlas Stream Processing and the Atlas Managed MCP Server, are available now. Teams can get started with Atlas for free at mongodb.com/atlas.
MongoDB is headquartered in New York and counts more than 67,000 customers across industries, including roughly 75 per cent of the Fortune 100.
Last Updated on August 15, 2026 by Nick Ross
