Renki is an AI-powered assistant designed specifically for Finnish newsrooms. Unlike traditional chatbots, Renki understands the unique context and workflow of journalism, helping journalists save time on everyday tasks such as transcription, editing, fact-checking, and content recommendations.

Developed by a small team rooted in Finland’s journalistic tradition, Renki integrates seamlessly with newsroom systems and supports journalists in creating reliable, high-quality news.

When we set out to build Renki, our aim was never to create just another chatbot. Renki was conceived as an AI newsroom assistant that understands not only words, but also the context, constraints, and rhythms of journalism — and frees journalists from time-consuming meta work.

In other words: a helpful colleague, not a mere tool.

The idea originated from a simple but persistent problem: the Finnish language. Integrating a reliable spellchecker and editor into our content creation platform had proven difficult. So, we decided to build it ourselves. Around that time, OpenAI made the ChatGPT API available. With spellcheck and AI-assisted editing, Renki was born.

It took our four-person team just a month to go from idea to launch. We are proud of our ability to drive editorial AI development through real-life use cases and great ideas — not through massive teams or endless budgets.

Our company has 109-year-old roots deep in the Finnish countryside. Renki, the farmhand, is a positive and helpful figure in Finnish tradition. The name felt particularly fitting. We wanted Renki to be inviting for journalists with no prior experience with AI. As it turned out, early Renki proved to be the perfect gateway app for journalists to familiarise themselves with artificial intelligence.

Today, Renki utilises several different language models. Some are better suited for article writing, while others are better suited for data analysis. System prompts guide each model to serve newsrooms, marketing teams, or coders based on their specific needs. Non-AI tools also coexist and remain just as useful for users.

Integrating with newsroom systems

Renki connects directly to our content creation tools, allowing journalists to use it at every stage of the editorial process.

The usual workflow goes as follows: a journalist uploads an interview to Renki for transcription. Background information is added to the transcription, which Renki combines into a draft. The draft is then developed and fine-tuned in our CMS.

The article returns to Renki for spellcheck and editing. Renki suggests photos and relevant articles to link to the story. It helps craft compelling headlines for all channels and inspiring social media posts.

Renki also knows when to add a paywall, lets you insert your own video or audio, estimates the story’s click potential, and more.

Renki chats about ideas, reads data, and selects relevant stories from various news feeds. It scouts search engines to identify trending topics and serves as an analytics tool. From statistics to A/B testing and weekly reports, Renki helps guide the newsroom to better performance. Its features have resulted in at least 20% time savings.

Looking forward

The recent Digital News Report 2025 from the Reuters Institute highlighted that 75% of Finns trust the news. Maintaining that trust is our top priority. Going forward, Renki will remain an assistant; nothing is published without a journalist’s review.

Our newest addition to Renki is fact-checking. In collaboration with Factiverse, a Norwegian pioneer in fact-checking, we are developing a tool for real-time analysis of political speech. The challenge is to maintain public trust in political reporting using AI-powered speech recognition, database integration, and advances in machine learning. The project is part of the GAMI Incubator #Finland programme.

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