During the 2024 Olympics, a large Finnish broadcaster’s sports desk was producing an astonishing 200 articles a day — and they needed fast, reliable access to athlete stats, historical data, and competition schedules. Vertti Luostarinen, a researcher & developer at CasvuGen, lead the development of Olympics Bot, a custom AI tool built using retrieval-augmented generation (RAG). Unlike general chatbots that sometimes “hallucinate” facts, the solution pulled from a trusted database: Finland’s sports history, bios of national athletes, official rules for every sport, schedules, and even live coverage feed. This meant commentators could get instant, accurate answers without wasting time digging through archives.
The result? Far fewer factual slip-ups than expected — though the bot’s Finnish spelling could still use some training. While the system occasionally mis-rendered names or declined to answer due to language limitations, it proved that RAG-powered tools can transform how journalists handle huge troves of data.
It’s important to remain cautious about AI’s tendency to oversimplify information, journalism often depends on nuance and detail. But this demo shows the potential for AI to be a powerful assistant — one that speeds up research while keeping humans firmly in charge of editorial judgment.