At Google Cloud Next ’24, the tech giant made a suite of announcements related to new capabilities in its all-in-one AI and ML platform, Vertex AI.
Gemini Comes to BigQuery and Looker
Google has introduced Gemini in BigQuery, currently in public preview, which leverages AI to enhance user productivity and cost-effectiveness throughout the analytics process. This feature stands out for its ability to understand user business context through metadata, usage data, and semantics. Gemini expands beyond chat assistance, offering new visual experiences such as data canvas—a natural language-based interface for various data tasks, including exploration, curation, wrangling, analysis, and visualisation.
Additionally, the company has launched Gemini in Looker, now in private preview, allowing business users and analysts to interact with their data conversationally. This integration offers conversational analytics, report and formula generation, LookML and visualisation assistance, and automated Google slide generation. These capabilities will soon be integrated with Workspace to streamline access to data visualisations and insights.
Building Generative AI Apps with Database
In February, the company announced AlloyDB AI, which enhances AlloyDB for developing enterprise generative AI apps with PostgreSQL. The latest AlloyDB AI version includes improved vector capabilities, simplified access to remote models, and secure natural language support.
To streamline inferencing endpoint management, it launched AlloyDB model endpoint management, facilitating access to Vertex AI, third-party, and custom models. It’s available in AlloyDB Omni and will soon extend to AlloyDB on Google Cloud.
It also added two features to AlloyDB AI to provide flexible, accurate, and secure natural language experiences. It enables precise data querying using natural language akin to SQL and introduces a “parameterised secure view” for enhanced data security based on end-users context.
Alongside advancements to AlloyDB, the company unveiled Bigtable Data Boost, enabling high-performance, workload-isolated processing of transactional data. Similar to Spanner Data Boost, it allows the execution of analytical queries, ETL jobs, and ML model training directly on transactional data without disrupting operations.
Moreover, it has also introduced authorised views, enabling secure sharing of data across multiple teams, and Bigtable distributed counters for processing high-frequency event data like clickstreams directly in the database, facilitating real-time operational metrics and scalable ML features.
Vertex AI Agent Builder
Google has introduced Vertex AI Agent Builder, which brings together Vertex AI Search and Conversation products, along with a number of enhanced tools for developers. It makes it easy to augment grounding outputs and take action on the user’s behalf with extensions, function calling and data connectors. Vertex AI extensions are pre-built modules for linking LLMs with specific APIs or tools.
Vertex AI function calling enables users to define a set of functions or APIs, allowing Gemini to intelligently select the appropriate API or function and parameters for a given query. Vertex AI data connectors also help ingest data from enterprise and third-party applications like ServiceNow, Hadoop, and Salesforce, connecting generative applications to commonly used enterprise systems.