When Benoit Scherrer was 12 years old, he spent all his free time coding away on his computer in Grenoble, France. At that time, he couldn’t have known that he would go on to use cutting edge agentic artificial intelligence (AI) technology to upend medical imaging operations to improve patient satisfaction.
His startup, co-founded in Boston with fellow Harvard Instructor Robert MacDougall, improves radiologists’ ability to understand the effectiveness of their expensive MRI scanner equipment by analyzing, cleaning, and harmonizing scanner data as well as scheduling data — giving imaging providers a never-before accomplished but much needed understanding of the function of their complex machinery and improving patient outcomes.
Scherrer’s journey to heading an advanced medical tech startup in the United States began, of course, 3,000 miles away, where, as a child he was fascinated with tinkering with computers and etching his own circuit boards and programming microcontrollers with machine language.
At 15, Scherrer was the youngest finalist at the time for Prologin, France’s 36-hour national programming marathon. His prodigy status helped lead him into an Engineering Degree in Mathematics and Computer Science, where he ranked second in his class.
Upon obtaining his bachelor’s degree, he moved on to study for his doctorate at Grenoble-INP. Before agentic AI became mainstream, he completed his PhD in Applied Mathematics while achieving the Best PhD award for creating a Bayesian agentic AI approach to 3D brain MRI analysis — a stepping stone into a field that would shape his professional life for decades.
After his PhD, Scherrer made the decision to move to Massachusetts to begin his journey into radiology at Boston Children’s Hospital. While at Children’s, Scherrer became a Senior AI Scientist and also an Assistant Professor of Radiology at nearby Harvard Medical School.
Over more than a decade, Scherrer published more than 50 peer‑reviewed papers, earned multiple awards, and secured two U.S. patents (with additional filings pending). Much of his time during these years was spent in the halls and radiology labs at Children’s, and it is precisely there where he would have what he describes as a pivotal, “aha” moment.
While speaking one day with his mentors Simon Warfield, the Director of Radiology Research, and Richard L. Robertson, the former Chief of Radiology at Children’s, Scherrer realized it was very difficult to measure the efficiency of expensive MRI machines, which can cost between $1 million and $4 million apiece.
What’s more, the hospital’s mandate that year was to reduce sedation for patients undergoing MRI and other types of at-times-uncomfortable scans. Children’s had come up with a faster protocol to reduce sedation, but at the time, they had no easy way to track adoption of the protocol at scale to see if it was working.
This “operational black hole” for radiologists would become his obsession.
He reached out to his fellow Harvard instructor and colleague at Children’s, Robert MacDougall, who he’d worked with on a previous medical imaging startup.
The original aim for their first prototype was to track the adoption of the new sedation protocols. That quickly evolved into the broader aim of achieving visibility into the operational efficiency of medical imaging machines and processes based on the mountains of data that scanners generate daily.
After receiving seed funding from Boston Children’s Hospital’s Radiology Foundation, the pair launched TheBox, a prototype which analyzed raw MRI metadata on the fly to shed light on the operational black hole.
TheBox would evolve into what is today Quantivly after the pair secured a federal research grant in 2021. In 2022, they raised their first pre-seed round of venture capital funding — $1.7 million, from NINA Capital and OneWay Ventures. They followed that with a $3 million seed round in 2023 and have been going from strength to strength since, securing key commercial deployments.
Their ambitions grew along with their resources. As they were increasingly shedding light on the “operational black hole”, the Quantivly team turned their attention from unlocking operational data and visibility to turning that data into real-time, AI-powered insights.
Today, Quantivly operates as a unified data layer that streams every breadcrumb from MRI, CT, PET and other scanners, RIS, and PACS into a single, queryable ontology.
That data layer has enabled the company to build and train its first foundational AI model for imaging operations which will help it learn and make sense of the workflow and operational efficiencies (and inefficiencies) of individual hospitals.
Currently in development, Quantivly’s agentic AI platform proactively orchestrates scanner scheduling, staffing, reimbursement capture, quality management, and more for hospitals and labs. The aim is to free up clinicians time so they can spend it with patients by allowing the AI agent to solve the complex logistics of medical imaging.
Quantivly’s use of agentic AI harks back to Benoit’s award‑winning doctoral work in Bayesian agentic systems. From his time tinkering with circuit boards, to pioneering in AI, there’s a clear line in the Harvard assistant professor’s journey to re-imagining the medical imaging industry.
