MongoDB announces updates to power generative AI development

Cloud database provider MongoDB Inc. today announced several new updates at its developer conference MongoDB.local NYC that will make it easier for developers to build, run and deploy generative artificial intelligence applications.

The company announced new capabilities for MongoDB Atlas, a fully managed cloud database that can handle the complex task of deploying and scaling database systems, including real-time stream processing, search nodes for Microsoft Azure to optimize performance and edge server services to reduce complexity for managing data.

Atlas Stream Processing, now generally available, allows developers to take advantage of real-time data coming from sources such as internet of things devices, customer browsing data or customer feeds and handle dynamic changing conditions rapidly. It allows them to handle and adjust to changing business logic quickly and ingest it into AI models with less time and less operational overhead.

Atlas Search Nodes, now generally available, allow developers to isolate workloads to optimize performance and costs by putting workloads on core operational database nodes for Atlas Vector Search and Atlas Search. Using this new system, customers can reduce query times by up to 60% and run highly available generative AI and relevance-based searches at scale. Atlas Edge Server, now in public preview, provides a local instance of MongoDB that can synchronize on local and remote infrastructure, allowing it to operate even on systems that have intermittent connectivity.

“Customers tell us they love MongoDB Atlas because it provides an integrated set of capabilities on one platform that can store and process their organization’s operational data across all of their applications,” said Sahir Azam, chief product officer at MongoDB. “The additional services we’re launching today for MongoDB Atlas not only make it easier to build, deploy and run modern applications but also make it easier to optimize performance while reducing costs.”

MongoDB Atlas Vector Search comes to Amazon Bedrock

Developers can now use MongoDB Atlas Vector Search Knowledge Bases integrated with Amazon Bedrock’s fully managed library of generative AI foundation models.

Bedrock is a service from Amazon Web Services Inc. that provides access to foundation models from numerous providers, including AI21 Labs, Amazon.com Inc., Anthropic PBC, Cohere Inc., Meta Platforms Inc., Mistral AI and Stability AI Ltd. Combined with Atlas Vector Search, customers can build large language models using company data converted into vectors for customization without the need for building a new model. Similarly, models can be enhanced to respond with relevant, real-time information for retrieval-augmented generation to reduce hallucinations.

Examples of this use could include a retail organization building a generative AI agent for customer service to process inventory requests and simple customer returns and exchanges without the need for a human agent. Having an up-to-date knowledge of inventory to suggest an exchange, especially if the item is out of stock, can speed responses and enhance customer experience.

“For more than a decade, AWS and MongoDB have been helping organizations transform their businesses with their data,” said Vasi Philomin, vice president of generative AI at AWS. “Today, tens of thousands of organizations choose Amazon Bedrock to build generative AI applications that are tailored to their specific needs.”

Google collaborates with MongoDB on Gemini Code Assist

MongoDB and Google LLC’s cloud division Google Cloud today announced a collaboration that will optimize Google Gemini Code Assist, an AI-powered coding assistant so that it can provide enhanced suggestions for developers building applications for MongoDB.

Code Assist is an enterprise-grade AI assistant that uses Google’s most powerful AI model, Gemini, to provide suggestions and answer questions about coding projects. The assistant works with the context of code in developers’ integrated development environments, tools that coders use to build, test and design applications before deploying them. It can also update entire codebases, explain code and help develop entire new functions based on questions to accelerate application development.

“Collaborating with Google Cloud to integrate Gemini Code Assist with MongoDB libraries and best practices will give developers the ability to build more quickly and to focus on more difficult tasks like ideating new types of application experiences for customers,” said Andrew Davidson, senior vice president of product at MongoDB.

With this collaboration, Code Assist will have a greater understanding of MongoDB, which will mean that it can provide more advanced techniques for dealing with complex situations. That’s especially so for those that might arise when developers are dealing with generative AI application programming. It also means that it will more easily answer common questions about MongoDB and deal with detailed database use cases in code, reducing tedium.

The new optimization for Code Assist will become available in the next few months, the two companies said.

Image: MongoDB