Phaze Ventures is pleased to share that portfolio company Sully.ai has been featured by NVIDIA for its use of open-source models and NVIDIA Blackwell infrastructure to scale healthcare AI agents.
The NVIDIA feature highlights Sully's work with Baseten to improve the economics and performance of healthcare AI agents at scale. According to NVIDIA, Sully reduced inference costs by 90%, improved response times by 65% for critical workflows such as generating medical notes, and has returned more than 30 million minutes to physicians.
In healthcare AI, infrastructure economics matter. Cost and latency directly shape what can be deployed, how widely it can be used, and whether an AI workflow can produce clear ROI for healthcare organizations.
Sully builds AI medical employees that support routine and time-consuming healthcare workflows such as documentation, medical coding, intake, triage, and administrative work. As the company scaled, the challenge became not just building useful agents, but running them with the reliability, speed, and cost structure required by real clinical environments.
NVIDIA's case study describes how Sully uses Baseten's Model API to deploy open-source models, including gpt-oss-120b, on NVIDIA Blackwell GPUs. Baseten combined Blackwell with low-precision NVFP4, TensorRT-LLM, and NVIDIA Dynamo to improve throughput and reduce cost per token.
For Sully, this infrastructure work translated into a major operational improvement: lower inference costs, faster response times, and more capacity to serve healthcare teams with AI agents that operate inside day-to-day workflows.
In healthcare, product quality and economics have to work together. Reducing inference cost while improving response times gives Sully more room to deliver measurable ROI for doctors and healthcare organizations. NVIDIA's recognition reflects the depth of the team's execution.
Abdullah Al Shaksy, Co-founder and CEO of Phaze Ventures
We first invested in Sully because we believed healthcare would become one of the most important markets for agentic AI. The company has continued to execute on that thesis, moving from early workflow automation toward a broader AI workforce layer for healthcare organizations.
The NVIDIA feature shows that Sully is not only building at the application layer, but also making the infrastructure choices required to scale AI in a sector where speed, reliability, cost, and trust all matter.
Sully's mission is to make doctors superhuman and help healthcare teams spend more time on patients instead of paperwork. Returning more than 30 million minutes to physicians is a meaningful step toward that goal.
We are excited to continue supporting the Sully team as they scale AI medical employees across healthcare.
About Sully.ai
Sully.ai builds AI medical employees for healthcare organizations. Its platform supports workflows such as AI reception, triage, scribing, medical assistance, and coding, helping providers automate routine work and spend more time on patient care.
Visit sully.ai to learn more.