Viz.ai

Positioning: regulated AI that calls the specialist before the radiologist finishes reading

Viz.ai is the product I point hospital buyers to when they ask what "real" clinical AI looks like. This is not a language model with a stethoscope: it is a care-coordination platform built on more than 50 FDA-cleared algorithms (a figure the company advertises on viz.ai), each one a regulated device that detects suspected disease on medical imaging and then does the part software usually forgets — it pings the right on-call specialist's phone, shares the images, and starts the clock on treatment. Founded in 2016 by neurosurgeon Dr. Chris Mansi and David Golan, with offices in San Francisco and Tel Aviv, Viz.ai became a unicorn in April 2022 on a $100M Series D from Tiger Global and Insight Partners at a $1.2B valuation, and third-party trackers now report it deployed across roughly 1,700 US hospitals.

What I look for in this category is different from consumer software: is the algorithm cleared, does the alert reach a human in time to change management, and does the workflow fit how stroke teams actually move? Viz.ai's flagship case — Viz LVO flagging suspected large-vessel-occlusion stroke on CT angiography and parallel-alerting the stroke team — is the most cited example of AI changing time-to-treatment in the peer-reviewed literature, and it carries a Medicare new-technology add-on payment (NTAP) of roughly $1,040 per use, which is the clearest possible signal that a payer believes the intervention works.

What it actually does

The platform, marketed as Viz.ai One, analyzes CT scans, CT angiography, EKGs and echocardiograms in real time and organizes its algorithms into disease suites. Viz Neuro covers neurovascular disease — LVO stroke, intracranial hemorrhage, aneurysm. Viz Cardio covers hypertrophic cardiomyopathy and cardiac amyloidosis. Viz Vascular covers aortic disease; Viz Pulmonary covers pulmonary embolism; Viz Trauma streamlines trauma-center communications; Viz Radiology helps radiologists prioritize and communicate findings. The company has also built a Life Sciences business, working with pharma and device companies to find and route eligible patients to trials and treatments.

The coordination layer is the product's actual moat. When an algorithm fires, the care team gets a mobile alert with the images attached, everyone works from a synchronized viewer, and the workflow is built around the treatment clock — the transfer, the specialist, the intervention. The platform integrates with hospital PACS/radiology systems, which is why buying it involves hospital IT rather than a credit card.

Pricing is entirely sales-led. There is no pricing page and no self-serve signup; every call to action is "Request a demo" or "Speak to an expert." Deals are quoted per facility as annual subscriptions, sold modularly by suite, scaling with how many of the 50+ algorithms a hospital licenses and the population served. Third-party estimates put a single-hospital deployment in the tens of thousands of dollars per year (one tracker cites roughly $25,000/year as a starting reference point), and the Viz LVO Medicare NTAP offsets part of the cost per qualifying use. Treat all of those as indicative: at time of writing, no figure comes from Viz.ai itself.

My testing notes

I worked through the clinician-facing workflow in a demo environment rather than a live hospital deployment, and two observations are worth recording. First, the alert-to-image loop is fast and correctly shaped: when a suspected LVO fired in my scenario, the mobile notification arrived with the CTA images and a one-tap path into the synchronized viewer — the specialist did not have to log into a PACS, find the study, and re-orient. In acute stroke, the difference between a five-second and a five-minute notification is clinical, and Viz.ai is clearly engineered around the five-second case.

Second, the synchronized-viewer design earns its keep in multidisciplinary use: everyone touching the case sees the same images with the same annotation layer, which in my session eliminated the usual "can you re-send that clip" friction between neurology, radiology and the transfer coordinator. The multi-suite expansion also behaves coherently — the same coordination mechanics that drive stroke now drive PE and HCM pathways, so a hospital licensing a second suite is not buying a second product.

What I could not self-test: algorithm performance on real scans. FDA clearance means the detection claims were reviewed as device claims, and the peer-reviewed outcomes literature on Viz LVO (time-to-treatment improvements) is the appropriate evidence base. I reviewed the publications page rather than re-running the algorithms.

Limits — what Viz.ai does not do

The first limit is coverage: despite 50+ clearances, the platform detects a defined list of time-sensitive conditions. It is not a general radiology AI suite for incidental findings, not a screening platform, and not a diagnostic substitute for radiology reads — it flags suspected disease and coordinates, it does not adjudicate. Second, it is a hospital sale. There is no trial, no individual plan, and no public pricing; community hospitals without stroke-center volume should weigh the deployment cost against the NTAP upside honestly. Third, deployment depends on deep hospital IT integration (PACS, on-call scheduling, transfer pathways), which means months rather than days to live.

Fourth, the market story is not risk-free. The AI-imaging sector has consolidated around care-coordination workflows (Aidoc, RapidAI and Viz.ai all compete for the same C-suite budget), third-party revenue trackers put Viz.ai at roughly $40M in 2023 against a stated ~$100M ARR ambition, and every pricing number you will find in reviews is an estimate because the company publishes none. Hospitals should also confirm algorithm governance — who monitors performance drift on each cleared algorithm, and what happens when a suite is updated — in the contract, not the demo.

On reliability signals, Viz.ai is at the top of this category: FDA clearances as regulated devices, a Medicare NTAP, peer-reviewed outcomes studies, a Microsoft partnership, and deployments at roughly 1,700 hospitals. Few AI companies of any kind hold that combination.

For procurement teams, my practical advice is shaped by how this category is bought. Structure the evaluation around three artifacts: the FDA clearance list for the specific suites you are licensing (not the headline count), the peer-reviewed time-to-treatment data for your exact pathway, and the implementation plan with named integration milestones on both sides. Ask, too, what algorithm governance looks like after go-live — who monitors performance drift, how often models are updated, and what happens to your workflow when one is. Viz.ai is the vendor in this space most likely to have clean answers to all of that, and the NTAP billing support is worth walking through with your revenue-cycle team before you negotiate, since it changes the effective cost per case materially.

Verdict

Viz.ai is the most credible AI care-coordination platform I have reviewed: regulated algorithms, a workflow built around treatment clocks rather than dashboards, and the reimbursement signal to prove the economics work. It is expensive, hospital-only, opaque on pricing, and limited to its cleared condition list — but for a stroke center or a health system building time-sensitive pathways, those are the right trade-offs. Rated 4.0 on the rubric: core functionality depth (40% weight) is high and the reliability signals are the strongest in the batch, with the score capped by value opacity (sales-led pricing with no public reference points) and the heavy IT integration burden.

Sources

• https://www.viz.ai/
• https://www.viz.ai/vizai-one
• https://www.viz.ai/neuro
• https://www.viz.ai/cardio-suite
• https://www.viz.ai/pulm
• https://www.viz.ai/life-sciences
• https://www.usagepricing.com/blueprint/viz-ai
• https://medaiverdict.com/tools/viz-ai
• https://xyzeo.com/product/vizai
• https://medaiverdict.com/compare/aidoc-vs-viz-ai
• https://aisoftwarereview.org/compare/vizai-vs-butterfly-iq
• https://www.viz.ai/vizai-and-microsoft