Hey HN, we are Kugelaudio a voice AI startup from Berlin. We are releasing our first speech-to-speech model, Fluid. We achived a Tau-bench score of 84% comparible to other frontier S2S models in the space. We focused on making the experience as natural as possible. It has great understanding to decide on turns, backchannels (like mhm, hm), interruptions and can also handle many tasks at ones and update you on long running ones. Tool calls are delgated to a frontier model to decouple intelligence. You can try it out on our platform.
Hey HN, we are Kugelaudio a voice AI startup from Berlin. We are releasing our first speech-to-speech model, Fluid. We achived a Tau-bench score of 84% comparible to other frontier S2S models in the space. We focused on making the experience as natural as possible. It has great understanding to decide on turns, backchannels (like mhm, hm), interruptions and can also handle many tasks at ones and update you on long running ones. Tool calls are delgated to a frontier model to decouple intelligence. You can try it out on our platform.