Willow Care Inc., the maker of the Willow Voice dictation app, announced it will use CoreWeave Inc.’s reinforcement‑learning infrastructure for training its models. The partnership lets Willow skip the hardware and software details and concentrate on evaluating alignment and post‑processing. Willow fine‑tunes a speech recognizer on quiet real‑world audio and then runs a tiny large language model to clean up the text. The move comes as the company expects dictation demand to grow faster than training workloads.
Why it matters
Willow can speed up improvements to its voice‑to‑text service without managing complex compute resources.
