Rime has launched Prism, a customizable speech-to-speech AI model for enterprise customer conversations. The San Francisco-based company says Prism was trained on millions of hours of human conversations. Businesses can fine-tune the model and deploy it within their own infrastructure.
Results from customer testing
In a trial with Revmo, Rime compared a fine-tuned Prism model with an OpenAI speech model. Prism achieved 100% on order confirmations and pickup methods, compared with 68% for OpenAI. Rime also reported higher rates of calls without hallucinations, at 85% versus 35%, and without dead air, at 88% versus 66%.
Rime's enterprise customers
Founded in 2022, Rime develops AI voice models for business conversations. Its customers include Asurion, Mayo Clinic, Upstart, and Dialpad. The company says its technology supports more than 100 million interactions each month across healthcare, finance, and customer service.
How Prism works
Prism generates text before speech, allowing businesses to check responses against internal policies. Companies can customize vocabulary, pronunciation, response speed, and safety controls. The model also detects conversational cues such as frustration and hesitation to support customer requests and call routing.
Prism is available for businesses to try through Rime's website. Companies can run the model on their own servers or in private cloud environments while retaining control over their data. Dialpad, which helped develop Prism as a design partner, plans to introduce the model across its AI agent products.