Butterfly Network has launched Compass AI, the latest generation of its enterprise software platform. It’s powered by artificial intelligence (AI) to reduce workflow friction and support scalable, revenue-ready point-of-care ultrasound (POCUS) programs.

The company said most POCUS programs still use radiology-style, order-based workflows, which can cause delays in documentation or orders being skipped altogether. Compass AI replaces that with an AI-enhanced, encounter-based workflow built for POCUS, allowing documentation compliance rates of up to 94% without adding work.

Traditional workflows capture about 15% of compliant studies, meaning that up to 85% go unbilled. Compass AI can translate that into as much as five times more revenue from the same clinical volume, the company said.

Compass AI works across departments and ultrasound devices, transforming POCUS from a set of siloed tools into a coordinated imaging service. It gives health systems the visibility and governance needed to scale, allowing administrators to track usage, credentialing, quality, and potential ROI—strengthening profitability, compliance, and care quality in real time.

A documentation agent uses ambient voice dictation to complete notes during exams, cutting charting time by up to 25%. A QA agent streamlines QA through AI-powered image review and feedback to reduce manual work and accelerate turnaround. A program dashboard tracks usage, compliance, credentialing, and unbilled scans.

It connects with EHRs, PACS, and SSO to route studies into the medical record and prevent lost images. Fleet visibility supports governance across users and devices for system-wide scale.

“Health systems want to expand POCUS and capture more value, but they’ve been held back by undocumented exams, compliance risk and administrative burden that take clinicians away from patients,” said Joseph DeVivo, chairman, president, CEO of Butterfly Network. “Compass AI changes that. It gives organizations the workflows and interoperability to manage POCUS programs at scale—no blind spots, no fragmented care, no added risk. To me, it’s a path to easy money: capturing value that was already there, just being missed.”

In September, the company touted its role in new research demonstrating the potential for machine learning models to support early aortic stenosis detection using handheld ultrasound devices.