To complement the Personal Agent Protocol, Sierra has developed the Fleming-1 model, which detects when the caller on the other side of a phone call is an AI https://t.co/MvjWln2zUO
Yesterday personal agent protocol. Today we're launching the latest in our constellation of Sierra models: fleming-1, which detects when a caller is another agent.
Read more: https://t.co/w4Gb6VxS8t
@btaylor@claybavor Curie powers the core orchestration loop of the conversation — deciding what path to follow, what tools to call, and writing the replies your customers see and hear.
@btaylor@claybavor Curie powers the core orchestration loop of the conversation — deciding what path to follow, what tools to call, and writing the replies your customers see and hear.
Love seeing the meaningful progress on tau-voice! It is a non-trivial benchmark based on our real world experience separating agent performance in voice from chat.
Our newest voice model is 20+ points ahead of the nearest competitor on Sierra's leaderboard. Pretty massive upgrade from our OG model from December, with significantly improved realism and tool calling capabilities.
@pitdesi One of the things I like about IIT entrance exams was that it’s based on global ranking of students on an objective score. This is obviously not perfect, but it avoids this inversion.
With Gemini CLI, and more recently the execution in AI from @Google has been just so amazing. It reminds me of a younger @Meta which would quickly copy and improve features launched by competition.
We often need to tweak emotions in voice agents. For example, I might need the voice to sound more empathetic or more excited, etc. if you have a system that can do this, I would love a demo!