Google unveils enterprise data science and engineering AI agents provide real-time analysis – News

In a bid to make the lives of enterprise data engineers and data scientists easier and developers easier, Google Cloud today announced the release of six new artificial intelligence agent tools.

The new tools include a new data engineering agent for BigQuery and a data science agent for BigQuery Notebooks, to assist across the data lifecycle from preparation to providing an intelligent workspace for data scientists for managing infrastructure, respectively. Business users will receive conversational analytics and a code interpreter that allows them to ask questions about data in natural language.

“To make this agentic reality possible, you need a different kind of data platform — not a collection of siloed tools, but a single, unified, AI-native cloud,” said Yasmeen Ahmad, managing director Data Cloud, Google Cloud.

For data engineers, Google is introducing the Data Engineering Agent in BigQuery in preview, which will simplify and automate complex data pipelines.

BigQuery is Google’s cloud-based data warehouse that allows data engineers to store, manage and analyze massive datasets using Structured Query Language, using a high-performance search at scale. BigQuery enables straightforward data analysis with familiar tools using standard SQL, a widely recognized database query language, making it accessible to many users.

Its architecture is also optimized for swift query execution, allowing for the analysis of vast amounts of data in seconds or minutes, providing easy integration into services such as Vertex AI for machine learning and Data Studio for data analytics.

Engineers can simply describe what they want done, such as “Create a pipeline to load a CSV file, cleanse the columns, and join it with another table,” and the agent will generate and build the entire workflow. The engineer can remain entirely hands off until the work is done and then modify it or fix it as they see fit, or iterate it with further prompts.

Google is also launching Spanner Migration Agent in preview, which will allow engineers to quickly move data from legacy databases such as MySQL.

Data scientists will get access to what Google calls a “reimagined” AI-first Colab Enterprise Notebook experience available in BigQuery and Vertex AI, featuring a new Data Science Agent in preview powered by Gemini, the company’s flagship AI model.

Colab notebooks, short for Google Colaboratory, are a cloud-based interactive environment that allows data scientists to write, execute and share Python code through a web browser. Combined with an AI agent, Google said users will be able to have the agent autonomously build entire analytical workflows, including exploratory data analysis, data cleaning, features and machine learning predictions.

The agent can plan, execute code and reason about the results as it proceeds before it presents its findings, just like another teammate, allowing the user to provide feedback and collaborate with it.

For business users and analysts

Last year, Google introduced a way for business users to “talk” to their data using natural language with the Conversational Analytics Agent. Today, the company is introducing Code Interpreter in preview.

Code Interpreter allows business users to provide complex natural language questions and transform them into executable Python code. This allows users with little or no technical knowledge to execute otherwise highly technical data analysis workflows and receive important insights.

“This enhancement supports the many critical business questions that go beyond what simple SQL can answer,” explained Ahmad. The agent generates code, provides explanations and creates interactive visualizations.

GitHub Actions now available in Gemini CLI

Google today announced an update to Gemini CLI, an open-source AI agent that brings the power of Gemini to the command line, which introduces a connection between developers and the GitHub repository.

The command line interface is the text-based way for developers to interact with a computer’s operating system. It differs from the easy-to-use drag-and-drop double-click icons of the desktop; instead, users must type out long strings of commands to get anything done. To ease this, Google’s Gemini CLI agent types them out for the developer.

Today, Google introduced Gemini CLI GitHub Actions in beta mode, which automates code repository actions, including pull requests, writing code, performing tests, providing review and implementation.

Google said the company is launching three open-source workflows that the agent will automate to help developers code faster. The new agent will help automate the overhead of managing new issues by helping analyze, label and prioritize incoming issues, to provide focus on what matters. It can also provide insightful feedback on code changes by reviewing them for quality, style and correctness. This will free up reviewers to focus on more complex tasks.

Finally, it will be possible to simply mention @gemini-cli within issues themselves to delegate tasks to the agent. For example, a developer has a feature they want implemented and submits it as an issue in the tracker, such as “Add a menu item to the Options drop-down that allows users to toggle the app to Dark Mode.”

The agent will then reply to the mention, begin work, make the code changes, generate unit tests and submit them to the repository for review. The developer can then pull them and check the build to make sure that the changes work properly and modify them as they see fit before committing them back, should they work or send them back to the agent for updates or modifications.

Google said these initial workflows are open-source and fully customizable, allowing developers to design and configure their own.

Images: Google