Reson8 co-founder Raoul Ritter told Slator that the company’s models “adapt to new vocabulary and context on the fly, just by providing text. No finetuning, no retraining, no latency hit.”
Reson8 is also leaning into its European focus, with a portion of the funding to go towards establishing a full Europe-hosted technology stack, including its own GPU clusters that it will use to train and run its own foundational speech models, said Ritter.
Beyond the physical infrastructure, Ritter said, “the bigger draw [to Europe] is the linguistic diversity. Languages like Dutch aren’t a priority for the big AI labs, but they’re core to what we do. Being here means we’re close to the problem, close to the customers, and close to strong engineering talent.” He said Europe’s stance on privacy and AI “aligns with how we’ve built Reson8.”
Reson8 currently supports more than 20 European languages, Balderton said in announcing its investment.
The company’s current offerings include a free tier that gets users five hours of streaming audio transcription and one custom domain model, and a EUR 100 per month plan for 300 hours of audio and 50 custom domains.
Ritter and his co-founders, Thomas Kluiters and Jarno Verhagen, all have a technical background, and Kluiters and Verhagen previously helped build AI medical transcription service Juvoly.
From here, Ritter said Reson8’s goal this year is to “give verticalized companies the best speech models for their specific use cases. For 2026, it’s about getting the product into more hands and proving that out.”
