Seed · AI
TypeSafe AI raises $40m seed for machine-native models
DCVC led the seed round for composable AI models that return typed decisions and confidence scores to software applications.
Models that return decisions applications can use
TypeSafe AI’s Jev model is built to classify, route and score inputs into developer-defined outputs rather than generate a free-form response. Applications can use its structured values and confidence scores to decide whether to automate a step or send it for review.
The company describes Jev as a System One model trained with reinforcement learning for calibrated decisions. Its machine-native outputs are intended to make semantic judgment a reusable component inside software workflows.
Use of funds
Expand operations and development efforts
Funding history
Seed · This round$40M