(Michael Vi/Shutterstock)
Google Cloud strengthens its analytical and transactional databases, including BigQuery, AlloyDB and Spanner, with new features designed to drive the development of generative AI applications among its customers.
BigQuery, which is Google CloudThe best database for powering analytics and AI workloads, has received several AI enhancements. First, the company rolled out a preview of an integration between BigQuery and Vertex AI for text and speech. This will allow users to extract insights from unstructured data such as images and documents, Google Cloud says.
Gemini, the company’s largest and most capable AI model – and which has also been the subject of some controversy following a rocky consumer debut last week – is also now available for BigQuery clients via Vertex AI.
These AI capabilities follow the previously announced vector search capability in BigQuery. The vector search feature, also previewed, enables critical components of GenAI applications, such as similarity search and retrieval augmented generation (RAG) using large language models.
Having access to Vertex AI directly in BigQuery enhances ease of use for Google Cloud AI customers in several ways, said Gerrit Kazmaier, general manager and vice president of data analytics.
“As a data analytics practitioner, you can access all Vertex AI models, including our Gemini (model) just from your SQL command line or the Python API built into BigQuery,” he said. Kazmaier said at a press conference yesterday. “It’s amazing because it means you don’t need to hire a data scientist or a machine learning platform. You can access it directly in the field you work in, directly on the data you have.
The second big benefit of integration is better access to data for AI models, Kazmaier said. Prior to this integration, getting data to AI models typically required building and operating as well as a data pipeline to move the data. This is no longer necessary, he said. “All that complexity disappears,” he said.
The ability to combine text and image-based AI models within Vertex – now available to data analysts through BigQuery – is also something that will greatly benefit customers, Kazmaier said.
“This opens the door to a whole new stage of analytical scenarios,” he said. “Synthesis, sentiment extraction, classification, enrichment, translation of structured and unstructured data. And it’s a huge deal. This is really the news here, because roughly 90% of the data available is unstructured. This data is generally not used in business data analysis because it is not possible to use it in a meaningful way.
On the transactional (or operational) side, Google Cloud announced the general availability of AlloyDB AI, the AI-specific version of the Postgres hosted database that the company unveiled at its Next 23 conference last year . Featuring the ability to store vector vector representations and perform vector search functions, Google Cloud sees AlloyDB AI as a critical component of its customers’ GenAI use cases.
Google Cloud also rolled out a new integration with LangChain, a popular open source framework that helps connect customer data into large language models (LLMs). All Google Cloud databases will be integrated with LangChain, said Andi Gutmans, general manager and vice president of databases at Google Cloud.
The new features were created in response to customer demand for a way to get more GenAI value from their data, Gutmans said.
“This is really what Gerrit and I dedicate our time to,” Gutmans said during the press conference with Kazmaier. “We have the data. We know that AI can’t succeed without data and so how do we make sure that it can actually work with data in concert and with real-time data.
The company also announced that it is adding vector search capabilities to other databases it hosts for customers on its cloud, including its Redis and MySQL offerings. Cloud Spanner, Firestore and Bigtable will also benefit from vector capabilities, Gutmans said.
“The unique thing about Spanner is that it will be an exact nearest neighbor search capability, which is a slightly different variation,” Gutmans said. “What’s really exciting is customers who have very, very large use cases, for example billions of vectors, highly partitioned based on users, for example. You can imagine that some of Google’s internal applications are sort of partitioned by user: they will be able to store and search vectors on the scale of a trillion (vectors).
All databases will eventually need vector functions, including the ability to store vector embeddings as well as some types of vector search functions, Gutmans said.
“Our belief is that any database, any place where you store operational data that you might need to use in a GenAI use case, should also have vector capabilities,” he said. he declared. “This is no different than 15 to 20 years ago when all databases added JSON support. We believe that good vector capabilities should simply maintain the fundamental capabilities of the database.
Related articles:
Google Vertex AI Search adds GenAI and enterprise-ready features
Google Cloud overhauls AI with Vertex launch
Google Cloud launches a new Postgres compatible database, AlloyDB
