Willow picks CoreWeave to run training and inference for its AI voice dictation app

The application combines speech recognition with language processing, according to Lawrence Liu, co-founder and chief technology officer of Willow. Liu described the arrangement in an interview with theCUBE Research's Dave Vellante and John Furrier at the Fully Connected event, during a broadcast on theCUBE, SiliconANGLE Media's livestreaming studio. The discussion covered CoreWeave's reinforcement learning infrastructure, speech recognition paired with a compact post-processing large language model, and Willow's goal of making keyboards obsolete.

"The great thing about CoreWeave's [reinforcement learning] infrastructure is we don't have to worry about anything under the hood," Liu said. "The only thing we really need to focus on at this point is eval alignment."

Willow fine-tunes its own speech recognition model on quiet, real-world audio, such as a user whispering in a coffee shop. Liu said speech recognition is only one part of producing usable text. "I think the main part of our secret sauce is we have this really, really small LLM that post-processes everything that you say," he said. "Now it's able to format everything correctly, take your inputs and change it to exactly what you would've written."

Most of Willow's work currently focuses on training, Liu said, though inference is growing faster as dictation demand climbs, a shift that aligns with CoreWeave's broader push into inference.

"Within two years, we want it such that you don't even need keyboards anymore," Liu said. "You can just speak and your computer understands."

TheCUBE is a paid media partner for the Fully Connected event. Neither CoreWeave, which sponsored theCUBE's event coverage, nor other sponsors had editorial control over content on theCUBE or SiliconANGLE.