When a veteran journalist needed to extract text from PDFs of local parade schedules, our AI Sandbox completed in seconds what would have taken hours manually.

When an Iraqi journalist in language training needed help with Norwegian grammar, the Sandbox let her focus on storytelling rather than syntax.

When reporters needed to compare this year’s municipal budget with last year’s, AI instantly revealed what had changed — and what had quietly disappeared.

These aren’t hypothetical scenarios; they’re every day wins at Amedia, Norway’s largest local news publisher with over 100 newspapers. While many newsrooms chase flashy AI experiments, we’ve taken a different approach: focus on practical tools that solve real journalistic problems.

Simple solutions, remarkable results

In February 2024, we launched our AI Sandbox, a secure environment where journalists could experiment with AI tools without technical expertise. We built straightforward functions for text improvement, title suggestions, SEO optimisation, and transcript generation — tasks that consume valuable reporting time.

The results speak for themselves: from 150 weekly users after launch, we’ve grown to 600 active users by December — more than half of our journalists. This wasn’t achieved through mandates or grand visions but through demonstrating immediate, tangible value.

The Norwegian approach to innovation

Our approach is distinctly Norwegian: practical, collaborative, and communal. We established an AI hub with representatives from eight different newsrooms, making them ambassadors and early testers. These journalists weren’t chosen for technical expertise but for their ability to spot opportunities where AI could solve real problems.

No one worked full-time on this project. Instead, developers, journalists, and editors contributed alongside their regular duties — what Norwegians call “dugnad,” our tradition of communal work for the common good.

From scepticism to daily use

“The sandbox has removed fear and increased understanding of AI in the newsroom,” said Marie Olaussen, development editor at Tønsbergs Blad. “We use it to improve language, sort datasets, plan interviews, transcribe recordings, and extract text from PDFs. We love the sandbox and use it every day.”

Pål Nisja-Wilhelmsen, development editor at Nettavisen, added: “As a journalistic tool, it’s like having a team of editorial assistants on standby at all times. Simply revolutionary.”

From simple tools to advanced applications

While initially focused on simple tasks, we're now exploring more advanced applications. When a local front page editor suggested developing an AI tool for Norway’s annual Tax List Day (when all citizens’ tax information becomes public), we built a natural language interface for analyzing 55 million tax records. The tool helps journalists uncover stories that would be impossible to find manually.

Key lessons

Our experience offers valuable lessons for newsrooms exploring AI adoption:

1. Simple wins build trust: Small practical solutions drive adoption more effectively than ambitious projects.

2. Let journalists teach journalists: Peer-to-peer learning outperforms top-down training.

3. Listen to the newsroom: Some of our best innovations came from journalist suggestions.

4. Start small, iterate fast: Launch with basic functionality and improve based on real usage.

The key to success wasn’t sophisticated technology; it was understanding how journalists work and meeting them where they are. By focusing on practical utility over technological wizardry, we’ve transformed AI from a perceived threat to an everyday newsroom tool.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .