From intuition to intelligence: Building a data-driven newsroom tool
Graciela Rock is the editor of La Cadera de Eva, part of the Mexican media outlet La Silla Rota. Learn how the JournalismAI Skills Lab helped her develop an internal tool that matches trending topics with audience metrics to help editors make smarter content decisions
At La Cadera de Eva – a news platform focused on gender, human rights, and social justice – editors were drowning in data yet making decisions by gut feeling. They had access to metrics and competitor information, but nothing connected. Weekly and monthly reports consumed hours to produce, only to be emailed and forgotten.
“Our editorial decisions were based on intuition,” Graciela Rock, the Mexican outlet’s editor, explains. “‘This sounds like something our audience would like,’ and ‘maybe this reporter can do it.’ It was not data-driven at all.”
The solution
With the help of the JournalismAI Skills Lab, a programme supported by the Google News Initiative, Rock built an internal tool using N8N that finds patterns between trending information and the newsroom's own metrics, identifying what will actually resonate with their audience.
The system pulls RSS feeds from other media outlets, runs them through AI agents for relevance scoring and sentiment analysis, then cross-references with Google Analytics, GA4 and Smartocto data. The output arrives as an email recommendation for editors.
What makes the approach distinctive is Rock's decision to break down the AI agent into separate, modular components. Rather than relying on one large agent, she isolated each function – geographical scope, sentiment analysis via Hugging Face models, author matching – so each element can be tuned separately.
“If I have every element that can be really tuned in to what they need and their tone and their interests, then it's really easy for me to adjust it for each vertical,” she says. The modular approach also reduces costs and makes the system more transparent.
Navigating challenges
Rock's journey wasn't linear. She arrived at the JournalismAI Skills Lab with an entirely different project – a historical documents chatbot – before pivoting twice based on instructor feedback. Her small team proved both advantage and challenge: decisions happened quickly via Slack messages but she also bore the burden of proof alone, spending weekends building before colleagues would commit time.
“If you're the one pushing for it, you're the one who has to put in the most hours,” she reflects. “Especially at first, because everyone's thinking, ‘Why are we doing this?’“
The turning point came when she demonstrated working results. Support materialised: authorisation for paid tool tiers, a colleague to assist, and genuine organisational buy-in. The team now plans to develop a visual dashboard and eventually offer the tool to other small Latin American newsrooms facing similar challenges.
Key takeaways
Hold space for curiosity. Rock emphasises that embracing experimentation means accepting frustration and mistakes. “You're going to forget to save and lose a week's worth of work,” she laughs. Curiosity requires patience with yourself.
Listen to those who know. Instructors and peers at the Skills Lab programme proved invaluable. Rock asked countless questions, pushed back when confused, and returned for explanations repeatedly. “Listen a lot,” she advises.
Trust your process. In workplaces without strong innovation cultures, you'll face indifference or scepticism. Rock learned to “armour yourself” with conviction. “This is good, this is important, I'm doing this because it's going to solve our problem” – such confidence helps you explain, advocate, and persevere.
Rock describes her confidence shift as dramatic: “Before, I didn't understand what was happening behind these AI tools. Now I better understand how it works, and how to educate myself further.” Her formula? “80% curiosity and 20% recklessness.”
