We built Radivault because report queues are not a radiologist problem. They are a systems problem.
The radiologist shortage is documented and not reversing on any timeline that helps today's department. Imaging volume has grown faster than reading capacity for the past decade. Radivault is the operational layer between the two: pre-reads to clear the backlog, triage to surface the urgencies, drafts to stop dictating from nothing.
Origin story
Samuel Okafor spent three years in clinical informatics at a Pittsburgh-area health network before starting Radivault in 2024. The work put him inside the gap between how radiology departments were supposed to operate and how they actually ran. Scan volume climbed every quarter. Reading staff did not. The PACS worklist had no intelligent order. Radiologists started every shift making the same triage decisions manually that could have been done before they logged in.
Every vendor solution he evaluated had the same problem: they were built for health systems with eight-figure IT budgets, or they required replacing the PACS entirely. Independent imaging centers and smaller department groups could not get to them. Radivault started from a different premise. Do not replace anything. Connect to what exists via standard HL7 and FHIR interfaces. Add the pre-reading layer. Get out of the radiologist's way.
Diana Chen, who had spent the previous two years building chest CT and plain-film classification pipelines, joined as CTO. Marcus Webb, with 12 years running HL7 and FHIR integration projects at regional health systems, joined to lead integrations. The three of them are still the team.
Founded 2024, Pittsburgh PA. Independently funded.
The team.
Samuel Okafor
CEO and Co-Founder
Three years in clinical informatics at a Pittsburgh-area health network, working on PACS integration and radiology workflow systems. That work surfaced the gap Radivault is built to close. Founded Radivault in 2024.
Diana Chen
CTO and Co-Founder
Medical AI researcher with a focus on image classification for chest CT and plain-film studies. Built and evaluated classification pipelines for two years before co-founding Radivault. Leads all model development and pre-reading output quality.
Marcus Webb
Head of Integrations
Twelve years in healthcare IT. Led HL7 v2 and FHIR R4 integration projects at three regional health systems, including PACS-to-EHR order and result messaging. Responsible for every customer integration at Radivault.
The operating principles.
Radiologist-first
Every feature decision starts with: does this reduce cognitive load or add to it? We observe workflows in the reading room, not just in product surveys. If a radiologist has to think about Radivault to use it, we have built it wrong.
Honest about what AI can do
Every output Radivault generates is labeled as decision support, not diagnosis. The impression block in a draft report carries an explicit [AI DRAFT, review required] tag. We design for the radiologist to be faster and to have the final word, not to be bypassed. Anyone who wants to use Radivault outputs without radiologist sign-off is using it wrong, and we say so plainly in our terms.
Start small, measure everything
Pilot on one modality. Measure TAT before and after. Pull the draft acceptance rate. Compare the backlog curve. Show the department the actual data, not a vendor deck. If the numbers support expansion, expand. If they do not, we want to know why before asking for a broader commitment.