Google Colab, Google’s cloud-based notebook tool for coding, data science and AI, acquires a new “AI agent” tool, Data Science Agent, which helps Colab users quickly clean their data, visualize trends, and gain insight into uploaded datasets.
Unveiled early last year at Google’s I/O Developer Conference, the Data Science Agent was first launched as a standalone project. However, according to Kathy Korevec, product director at Google Labs, Google has decided to integrate it with Colab with the goal of allowing users to access agents directly from Colab notebooks.
Data Science Agent is available for free on Colab as of this week, but Colab limits its free users to relatively small amounts of computing. Google offers a variety of paid co-love plans with a high limit starting at $9.99.
Data Science Agents are primarily targeted at data scientists and AI use cases, but agents can also help you spot API anomalies, analyze customer data, and write SQL code. All users need to do is upload the data and ask the agent questions.
Data Science Agent uses Google’s Gemini 2.0 AI model family in the backend and uses “inference” tools to assist with functional engineering and data cleaning tasks. Korevec told TechCrunch that Google is constantly improving its agents, using techniques including augmented learning, and integrating user suggestions to improve the performance of its data science agents.
Data Science Agent currently supports only CSV, JSON, or .TXT files in sizes less than 1GB. You can analyze approximately 120,000 tokens at one prompt. It works in about 480,000 words.
Korevec said data science agents could reach additional development-centric Google apps and services in the future.
“We’re scratching the surface of what people can do here,” she said. “It’s an agent, so you can integrate it into a bunch of different tools. I don’t necessarily want to force anyone who is embarrassed to look at the code and go to Colab.”