MongoDB adds vector search to Atlas database to help build AI apps

After trying to expand its user base to include traditional database professionals last year, MongoDB switched gears and rolled out features to turn its NoSQL Atlas Database as a Service (DBaaS) into a more complete data platform for developers. Adding. Build a generative AI application.

In addition to introducing Atlas’ vector search and integrating Google Cloud’s Vertex AI underlying model, the company announced at the MongoDB.local conference in New York on Thursday that it will showcase a range of DBaaS offerings, including the new Atlas Search, data streaming and query capabilities. We announced a new feature. .

“Everything MongoDB announced can be seen as a move to make Atlas a more inclusive and complete data platform for developers,” said Doug Heschen, principal analyst at Constellation Research. I’m here. “The more MongoDB can provide developers with all the tools they need, the more powerful the platform will be for them and the companies they work for.”

Henschen’s view seems reasonable given that the company competes with cloud data platform suppliers such as Snowflake, which offers native application frameworks, and Databricks, which recently launched Lakehouse Apps.

Vector Search Helps Build Generative AI Apps

In an effort to help companies build applications based on generative AI from data stored in MongoDB, the company introduced a vector search capability within Atlas called Atlas Vector Search.

According to the company, this new search capability will help support a new range of workloads, including semantic search by text, image search, and highly personalized product recommendations.

According to Matt Aslett, research director at Ventana Research, searches are performed on vectors, multidimensional mathematical representations of features and attributes of raw data such as text, images, audio, and video.

“Vector search takes advantage of vectors to perform similarity searches by enabling rapid identification and retrieval of similar and related data,” Aslett said, noting that vector search is a large-scale language model ( LLM) and can alleviate concerns about accuracy and reliability. Incorporation of Approved Corporate Content and Data.

Vector Search in MongoDB Atlas also allows companies to use open-source frameworks such as LangChain and LlamaIndex to augment pre-trained models such as GPT-4 with their own data, the company said. .

Using these frameworks, you can access LLMs from MongoDB partners and model providers such as AWS, Databricks, Google Cloud, Microsoft Azure, MindsDB, Anthropic, Hugging Face, OpenAI to generate vector embeddings and AI on Atlas. You can build applications that take advantage of it. Added.

MongoDB partners with Google Cloud

The partnership between MongoDB and Google Cloud to integrate Vertex AI capabilities aims to accelerate the development of generative AI-based applications. According to the company, Vertex AI provides the text embedding APIs needed to generate embeddings from enterprise data stored in MongoDB Atlas.

These embeddings can later be combined with PaLM text models to create advanced capabilities such as semantic search, classification, outlier detection, AI-powered chatbots, and text summarization.

The partnership will also enable businesses to get hands-on help from MongoDB and Google Cloud services teams in designing data schemas and indexes, structuring queries, and fine-tuning AI models.

Dremio, DataStax, and Kinetica databases have also added generative AI capabilities.

MongoDB’s move to add vector search to Atlas is nothing special, but it will give the company a competitive edge, Aslett said. “The list of specialist vector database providers is growing, and multiple vendors of existing databases are working to add support for introducing vector searches to data already stored on their data platforms,” said Aslett. said Mr.

Manage real-time streaming data from a single interface

To help businesses manage real-time streaming data from multiple sources with a single interface, MongoDB has added a stream processing interface to Atlas.

Called ‘Atlas Stream Processing’, the new interface can process any kind of data and has a flexible data model that allows businesses to analyze data in real time and tailor application behavior to end-customer needs. The company said it could.