Google upgrades Colab with AI Agent tool

Google Colab, a Google laptop, data science and AI, wins a new AI agent, a data agent, to help Colab users quickly clean data, visualize trends and get an idea of ​​their uploaded data sets.

First announced at the I/O Developer Conference on Google earlier last year, originally launched as A Independent projectS Google, however, has decided to integrate it into Colab in order to help users have access to the agent directly from Colab notebook, Katie Donevek, director of the Google Labs product, said in an interview.

The scientific data agent is available free of charge from this week in Colab, although Colab limits free users to a relatively low amount of calculations. Google offers a set of paid Colab plans with higher restrictions starting at $ 9.99.

The scientific data agent is primarily aimed at scientists according to AI data and cases, but the agent can also help find API abnormalities, analyze client data and write SQL code. All users have to do is upload their data and ask the agent.

The scientific data agent uses the Gemini 2.0 AI family of Gemini 2.0 at the back, along with the “reasoning” tools to help with the features of engineering and data cleaning tasks. Korevec told TechCrunch that Google is constantly improving the agent and uses techniques such as reinforcement training, as well as integrates users’ suggestions to improve the efficiency of a data agent.

Currently, the scientific data agent currently supports only CSV, JSON or .TXT files below 1 GB of size. It can analyze about 120,000 tokens with one prompted, which works up to about 480,000 words.

Korevec said the data agent can reach additional Google Apps aimed at Google Apps and services in the future.

“We scratch the surface of what people can do here,” she said. “As this is an agent, we can integrate it into a bunch of different tools and I do not necessarily want to force people who are ashamed to watch the code to go into a collapse.”