Medicare used AI to pre-approve patient care, and it immediately led to a large drop in medical services
THE SO WHAT
An AI-driven prior authorization system that immediately cuts medical services is a policy lever as much as a tech deployment—utilization management is being automated. Health systems and clinicians need to understand these models’ criteria now, or they’ll be reacting to opaque denials at scale.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIMistral releases open weights for its 1 trillion-parameter Large 4 model
Open‑weight access to a 1T‑parameter model that’s been sharpened with three extra weeks of RLHF and customer feedback compresses the gap between proprietary and self‑hosted top‑end capability. If you’ve been blocked by closed APIs for compliance or latency reasons, it’s time to re‑run the build‑vs‑buy analysis on your highest‑value workloads.
Applied AIThis $7,000 RTX Spark laptop had no business beating my RTX 5090 — but it did
A $7,000, 128GB unified‑memory laptop outpacing an RTX 5090 desktop on memory‑heavy AI workloads shows how fast “developer‑grade” is moving to portable form factors. For teams building local AI tools or field‑deployed inference, device constraints are loosening—your bottleneck shifts to power, cooling, and data access, not raw compute.
Applied AIBridgewater CEO Nir Bar Dea on AI Investing, Markets and Leadership | The David Rubenstein Show
When Bridgewater’s CEO frames AI as core to the future of investing, that’s a signal that systematic funds will treat model and data pipelines as primary edge, not back‑office tooling. Asset managers who haven’t yet built an internal AI thesis and experimentation sandbox are ceding narrative—and eventually flows—to those who have.
Applied AIVitVio brings its 3D operating room AI to a hospital in Mexico
Rolling VitVio’s 3D OR AI across every operating room in a Mexican hospital is a step toward computer‑vision‑driven workflow control in surgery, not just decision support. Hospital operators should start treating cameras and spatial models as core infrastructure—and get ahead of the consent, liability, and integration questions before scale arrives.