Blog
- Case study
Evaluating a form-filling voice agent that enrols mothers over a phone call
Our evaluation design for a voice agent that fills a form over a phone call, built on simulated calls, and what the first results say about the models we can deploy.
Calibrate will be at IndiaFOSS
Our proposal to present Calibrate at IndiaFOSS, the biggest gathering of the open-source community in India, has been accepted.
Evaluation is all you need
Evaluation is the missing piece that translates model capability to real impact.