Founded in 1919, Brazilian newspaper Gazeta do Povo prides itself on thorough coverage of politics and the economy. “We’re recognized for producing articles based on in-depth contextualization, not scoops,” says Director of News Ewandro Schenkel. As the oldest and largest newspaper in the state of Paraná, and in the top three countrywide for digital subscribers, Gazeta has survived for over 100 years by adapting to readers’ evolving interests and habits while using technology to help deliver timely, relevant content.

In 2017, Gazeta became the first mainstream Brazilian news outlet to transition to a fully digital daily format. A popular weekly magazine is also available as a PDF exclusively for subscribers. To make the most of their digital-first status, Audience Coordinator Bruna Maestri has been collaborating with the Google News Initiative, which gave Gazeta staff 13 training sessions on AI and other Google solutions in 2025 alone. “We wanted to identify opportunities to improve engagement and develop a strategy to deliver content more effectively,” explains Maestri.

NotebookLM, Google’s AI research tool, proved especially popular, with 85 team members learning to use it. Due to the in-depth nature of Gazeta’s reporting, journalists have to sift through a large volume of source material daily, from complex scientific papers to 300-page legal rulings. “We were delighted to learn that we could upload a document, get a summary, and ‘interview’ it by asking questions to extract information, instead of taking hours to read through it,” Editor-in-Chief Jones Rossi recalls. The team also appreciates that NotebookLM draws only from their uploaded sources and provides citations to back up its answers. “It enables us to verify information and transform it into news stories with far greater efficiency,” says Schenkel. “This represents a massive reduction in turnaround time and a significant boost in production agility.”

For Carlos Coelho, editor for subscriber strategy, NotebookLM’s value lies in its data analysis support. “We have so much information it’s difficult to understand the patterns,” he says. “We wanted to turn raw data into achievable insights and help our teams understand what works — or doesn’t work — for our audience.” So Coelho uploaded engagement metrics into NotebookLM and created an AI agent that can provide insights on topic performance, headline creation, and more, helping staff produce articles that resonate with readers.

Since the initial training, the team has expanded their use of NotebookLM. For example, Rossi uploaded Gazeta’s style guide to the tool to quickly check that articles follow editorial standards. They’re also using it to summarize long articles, which has helped increase engagement. And Maestri uses NotebookLM to summarize meetings, record decisions, and identify next steps, leading to clearer responsibilities and better accountability.

The team emphasizes that they thoroughly test the technology, are completely transparent about its use, and verify AI-generated insights with external programming tools. Schenkel also makes clear that they aren’t using AI to generate new articles, but rather to blend existing content and speed up research and data analysis for increased productivity. “We no longer have room to debate whether or not to engage with AI,” he says. “Our mindset is to transform our content so it’s presented in the most reader-friendly way possible.”