Willow Care Inc., the developer of the Willow Voice dictation application, has selected CoreWeave Inc. to support its AI model training infrastructure. According to Willow co-founder and CTO Lawrence Liu, the partnership allows the startup to focus on model evaluation and alignment rather than managing underlying compute resources.
The Willow Voice application integrates speech recognition with a compact large language model for post-processing. This system is designed to format spoken input into written text, with the company fine-tuning its speech recognition model on real-world audio samples, such as ambient noise in public spaces.
While Willow currently prioritizes model training, the company reports that inference demand is increasing. This trend aligns with the strategic focus of CoreWeave on expanding its inference capabilities. Willow aims to enable voice-based computer interaction that could eventually replace the need for keyboards within two years.
