How VG Built an AI “Buddy” for Journalists

# News Organizations

# EMEA

# Norway

# Advanced Tools for Newsrooms

VG’s Buddy Reader acts as an editorial sparring partner, helping reporters refine stories, reduce repetition, and simplify language within existing newsroom workflows.

This resource was produced by VG - owned by the media company Schibsted, a trusted partner in Norway. While OpenAI Academy shares it here, it does not necessarily represent OpenAI’s own views.

—

VG, Norway’s largest newspaper and the country’s leading digital news publisher, has built more than 30 internal AI tools, many of them powered by OpenAI’s API. One of the most popular is also one of the simplest: “The Buddy Reader”, a tool designed to help journalists improve stories before publication.

The Buddy Reader is not there to replace reporting or editorial judgment. It helps with the parts of writing that often take time in a busy newsroom: improving clarity, cutting repetition, correcting grammar, tightening structure, simplifying language, and explaining difficult words in a way readers can understand.

Journalists at VG are interested in new technology, and always have been, but most did not need another flashy AI demo. They needed something useful in the middle of real work. Something that could help when a story was nearly finished, but still needed feedback. Yes, their editors also read through stories, but they are busy too, so we needed a tool that could make the news stories better.

The Buddy Reader works like an editorial sparring partner. A journalist drops in a draft, either through VG’s internal AI toolbox site or inside the CMS via AI Playground, and the tool helps improve the text. It can spot language that is too complicated, point out repetition, suggest a clearer structure, and highlight where a word may need more explanation for readers, along with an explanation.

One reason the tool has worked well is that its value is easy to understand right away. It solves a real problem in a real workflow. It is also flexible. Journalists can choose between different “buddies” with different roles, depending on what they need help with, from headline and front-title improvement, word explanations, repetition detection and grammar correction to fact boxes, polls, quizzes and audience adaptations such as writing for younger readers.

We can also easily add new “buddies” with different roles and strengths, depending on what journalists need help with. One example was a VG+ buddy, designed to give better feedback on premium story ideas and help sharpen the value proposition of articles meant to drive conversion and subscriptions.

We keep trust in the tool by making sure a journalist checks everything before publication. The Buddy Reader is there to help, not to decide on its own, so all AI suggestions are reviewed by a person before they are used. This is especially important for things like explaining complex words, which must be checked by the journalist before the story get published.

The Buddy Reader says something larger about how we work with AI at VG. We are not trying to replace journalists. We are trying to build tools that save time, improve quality, and fit naturally into newsroom workflows, so journalists can spend more time on the reporting, judgment, and original work that AI cannot do for them. The Buddy Reader now handles around 400 queries per week, and roughly half of the newsroom uses the tool in a typical week.

That is why The Buddy Reader matters. It is a simple tool, but it solves real problems. And in a newsroom, that is often what matters most.

Table Of Contents

How VG Built an AI “Buddy” for Journalists

VG’s Buddy Reader acts as an editorial sparring partner, helping reporters refine stories, reduce repetition, and simplify language within existing newsroom workflows.

Popular

Related