a2z Radiology AI, a Boston-area startup founded in 2024, closed a $4.5 million seed round on December 9, 2025, led by Khosla Ventures and SeaX Ventures, just weeks after the FDA cleared its flagship product, a2z-Unified-Triage, on November 26, 2025. The device is the first AI system cleared to simultaneously flag seven urgent findings from a single abdomen-pelvis CT scan, and the company says commercial rollout is planned through 2026.
What the device flags
a2z-Unified-Triage screens for small bowel obstruction, acute cholecystitis, acute pancreatitis, acute diverticulitis, hydronephrosis, free air, and unruptured abdominal aortic aneurysm — all from one CT interpretation pass. According to the company, five of those seven conditions have never before had an FDA-cleared AI triage tool in the U.S. market, meaning radiologists previously had no automated first-pass flagging for those specific emergencies on abdomen-pelvis imaging, unlike more mature AI categories such as stroke or pulmonary embolism detection.
The people and the pitch
a2z was co-founded by Pranav Rajpurkar, an associate professor at Harvard Medical School known for his machine-learning research on medical imaging, and his brother Samir Rajpurkar, who serves as CEO. “The response at RSNA validated what we’ve been building — AI that considers everything,” Pranav Rajpurkar said, referring to the Radiological Society of North America’s annual meeting, radiology’s largest industry gathering. Vinod Khosla, whose venture firm led the seed round, said in a statement that “radiology is one of the areas that will benefit the most from AI, ensuring no disease goes undetected.” Coverage of the round has framed a2z’s pitch as part of a broader industry shift away from single-condition “point solutions” toward comprehensive AI systems designed to replicate a radiologist’s holistic read of an entire study, rather than hunting for just one abnormality at a time.
Why abdomen-pelvis CT is the target
Abdomen-pelvis CT is one of the highest-volume imaging studies in American emergency medicine, with more than 20 million exams performed annually in the U.S., according to figures cited by the company. Each scan can contain any one of dozens of possible acute findings, and radiologists reading emergency-department studies are under constant pressure to avoid missing a life-threatening finding buried in a complex scan. In clinical testing presented around RSNA’s 2025 annual meeting, a2z said its tool produced a 17.8% reduction in reporting time, a 14.8% increase in radiologist confidence, and a 22.4% decrease in self-reported mental workload among readers using the tool, while improving detection of the targeted findings without increasing false-positive flags.
A market already filling up fast
a2z enters a radiology AI triage market that already includes larger, better-funded players. Aidoc, one of the category’s pioneers, has FDA clearances spanning stroke, pulmonary embolism, and intracranial hemorrhage detection, while GE HealthCare’s 2025 acquisition of Intelerad signaled that imaging incumbents are racing to bundle AI triage into enterprise picture-archiving systems. Rad AI, a separate radiology AI company focused on report generation, raised $68 million in a Series C round backed by health systems including Advocate Health and Memorial Hermann. a2z’s bet is that consolidating seven distinct triage algorithms into a single FDA-cleared device, rather than selling point solutions one condition at a time, will be more attractive to hospital IT departments trying to avoid stitching together a dozen separate AI vendors.
The broader radiology workforce backdrop
The launch comes against a well-documented national shortage of radiologists, with hospital systems and outpatient imaging centers across the U.S. reporting growing backlogs of unread scans and lengthening turnaround times for emergency-department studies. Industry groups have pointed to rising imaging volumes, an aging radiologist workforce nearing retirement, and a comparatively slow pipeline of new residency graduates as structural pressures that AI triage tools are increasingly being asked to help offset, by helping the radiologists who are available work through studies faster and flag the most urgent cases first rather than reading scans strictly in the order they arrive.
Skeptics want more data
Radiology AI adoption has repeatedly outpaced independent validation, and some radiologists caution that vendor-reported confidence and workload metrics, gathered in controlled testing environments, do not always hold up once a tool is deployed across diverse hospital systems with different scanner hardware, patient populations, and radiologist experience levels. A $4.5 million seed round is also modest by health-tech AI standards — a fraction of the $60 million-plus rounds raised by more established radiology AI companies — leaving open the question of whether a2z has the capital to fund the multi-site, peer-reviewed studies that typically follow FDA clearance and drive broader hospital adoption.
What’s next
a2z says the seed funding will go toward scaling its commercial deployment through 2026, and the company is expected to pursue additional FDA clearances covering other CT body regions beyond the abdomen and pelvis. Whether health systems adopt a2z-Unified-Triage at scale will likely hinge on real-world evidence generated in the next year, along with how the startup prices itself against both point-solution rivals like Aidoc and the bundled AI offerings increasingly built into imaging platforms from GE HealthCare and Siemens Healthineers. Investors will also be watching whether a2z can convert its FDA clearance into signed contracts with hospital radiology departments before larger, better-capitalized competitors expand their own product lines to cover the same seven conditions.
