
Creating quality training data remains a major bottleneck for building reliable AI agents that can handle business tasks.
AutoSynthData is a system that generates training data for enterprise agents by identifying what a model struggles with and creating new tasks to teach those capabilities. Enterprises need agents tailored to their specific environments, systems, and workflows, but turning individual model failures into useful training data requires many realistic tasks that the agent cannot yet solve consistently. The system works by evaluating a target model's performance, using a stronger teacher model to understand what could be learned, and then automatically generating varied tasks with verifiers to check if solutions are correct. This approach creates a continuously improving curriculum where the system identifies remaining capability gaps and generates new training tasks accordingly.

A hospital giant and radiology network turned to Palantir to streamline scheduling, but nurses and other staff say the new software is causing errors, burnout, and frustration.

With all the talk about how AI might one day kill us all, it's easy to forget that AI has already harmed some people psychologically. Circuit Breaker Labs has created "crash-test dummies" to solve that.

Learn how startups can choose GPT-6 models, tune reasoning effort, improve prompts and skills, coordinate tools, and prepare workflows for production.
Want to go deeper than the news? Explore live, cohort-based AI courses taught by practitioners.
Browse AI courses on Maven