What Is RADPAIR? The Web-Based AI Radiology Reporting Platform Explained

The short answer: RADPAIR is a radiology reporting platform built around generative AI. Radiologists dictate naturally into a browser, and the platform handles punctuation, formatting and template population before drafting findings and impressions for review. It was founded in 2023 and is based in Knoxville, Tennessee, and Intelerad announced a prime partnership with it in May 2025. There is nothing to install. It is a documentation product: it does not read images, and the radiologist reviews, edits and signs every report.

Updated August 2026. Product capabilities in this category change quickly, so confirm current features and integrations with the vendor before planning around them.

RADPAIR started showing up on US radiology shortlists for a specific reason. PowerScribe 360 annual renewals and maintenance end on August 31, 2026, so a large slice of American radiology is being pushed into a reporting decision it did not schedule. Once you start looking, every vendor that touches the report appears in the same search results, and the names blur together fast. This guide covers what RADPAIR actually is, so you can work out whether it belongs on your list before you take a demo.

What is RADPAIR?

RADPAIR is the system a report gets written in. That single sentence separates it from most of the AI vendors around it. It is not a detection tool, it is not a triage platform, and it does not reorder your worklist. It is the place a radiologist dictates, and the generative models sit inside that dictation loop rather than bolting onto it afterwards.

The company describes the platform as completely web-based with no installation. You sign in and work. For a group used to a reporting client installed on specific workstations, with a configuration that has accumulated for a decade, that is a genuinely different operating model. It is also the reason the product turns up more often with teleradiologists and radiologists who read across multiple sites and systems, since a browser travels in a way an installed workstation does not.

What does RADPAIR do?

Three things, in one loop. It captures speech, so you dictate naturally and the platform applies punctuation and formatting rather than making you speak the commas. It generates, so instead of you narrating a full structured report, the models draft findings and an impression into the template shape you work in. And it lets you revise, through what the company calls Dynamic Editing, a click-and-drag approach to changing generated text rather than re-dictating a paragraph to fix a word.

Two named capabilities sit on top of that. PAIR INSIGHTS is the near real-time report generation piece. WINGMAN is described as an AI co-pilot. The vendor does not publish deep technical documentation for either on its public site, which is worth noting because it means both are demo questions rather than things you can evaluate from the outside.

How does RADPAIR generate a report?

The pattern is the one most generative reporting tools now use, and it is worth understanding because it changes what "accuracy" means. The radiologist reads the images and speaks the findings. The model turns that speech into structured, formatted report text and drafts the impression from the findings just dictated. The model is working on your words, not on the pixels. It has no independent view of the study.

That distinction matters more than it sounds. A generative reporting tool cannot catch something you did not see, and it cannot verify something you did. Its failure mode is a clean, confident, well-formatted paragraph that says something slightly different from what you meant, which is a harder error to spot than a garbled one. Every group evaluating this class of product should read a batch of generated drafts against the dictation that produced them before deciding how much editing time they actually save.

Who makes RADPAIR?

RADPAIR is a company of the same name, founded in 2023 and headquartered in Knoxville, Tennessee. Its CEO is Dr. Avez Rizvi, who framed the company's purpose in the Intelerad announcement as founded "with a mission to help radiologists thrive, by reducing burnout and bringing joy back to their work."

Company age is a real diligence item here rather than a snob point. A 2023-founded vendor is a different risk profile from a platform with two decades of installed base, and that cuts both ways. You get a product designed after generative models existed, rather than one retrofitted around them. You also get less certainty about who supports the software in five years, which is exactly the question US groups are asking about their current reporting platform right now.

Is RADPAIR partnered with Intelerad?

Yes. Intelerad announced a prime partnership with RADPAIR on May 20, 2025. The stated combination is Intelerad's workflow orchestration with RADPAIR's agentic AI, automating portions of the reporting workflow and pre-populating reports for review and approval. Intelerad is now a GE HealthCare company, which is relevant context if your imaging stack already sits in that family.

The announcement also carries the efficiency figure that follows RADPAIR around: early user feedback suggesting reporting efficiency gains "often by 30-40%" compared with traditional smart reporting solutions. Read that carefully before it lands in a business case. It is early user feedback rather than a published study, the comparison baseline is traditional structured reporting rather than nothing at all, and it is a vendor announcement. It is a reasonable reason to run a shadow period. It is not a number to budget against.

How much does RADPAIR cost?

RADPAIR publishes more about pricing than most of its competitors, which is unusual enough to be worth stating plainly. The company offers a seven-day trial with 20 free reports and no credit card required, then a pay-as-you-go credit model for individual radiologists. Enterprise licensing is quoted separately for private practices, teleradiology groups and health systems, through sales rather than a public price list.

Nuance PowerScribe and Rad AI both publish nothing. That asymmetry is genuinely useful to a buyer: it means you can put RADPAIR in front of a radiologist and get a real opinion before any contract, procurement cycle or scoping call exists. Very little else in enterprise radiology software allows that. Just be aware that per-report credit pricing and an enterprise agreement are different economics, and the trial tells you about the product rather than about your eventual bill.

Does RADPAIR replace PowerScribe?

It occupies the same layer, so yes in principle. RADPAIR is where the report gets dictated and written, which is exactly PowerScribe's job, and running both would be redundant. Whether it replaces it in your specific environment is a different question, and it turns on integration and templates rather than on which product demos better. We set the two against each other, along with Rad AI, in a full side-by-side table on RadPair vs PowerScribe vs Rad AI, including deployment, pricing model and the federal contract footprint behind each name.

The comparison worth understanding first is the layer one. Rad AI is not in the same layer at all: Rad AI generates the impression inside whichever platform you already run and tracks incidental-finding follow-ups afterwards. You can run it alongside PowerScribe. You would not run RADPAIR alongside PowerScribe. Buying two products that do the same job is the most common expensive mistake in this evaluation.

What RADPAIR does not do

It does not look at images. Nothing in the published material describes finding detection, and the models work on dictated text rather than on the study. If your problem is that something urgent sat unread in the queue for three hours, a reporting platform does not touch it. That is a worklist prioritization problem, and the acute end of it belongs to the acute triage platforms Aidoc, Viz.ai and RapidAI.

It does not publish follow-up tracking as a product area, which is the specific gap Rad AI Continuity fills. It does not claim FDA clearance, and it should not need to: a documentation tool that drafts text from your dictation is a different regulatory object from a device that flags a suspected finding on an image. Our guide to what FDA clearance actually tells you covers why vendors in this category say so little about it.

What to check before you buy

Four things decide this, and none of them is visible in a demo.

Integration comes first. Ask precisely what connects to your PACS and RIS, who builds that interface, and what it costs. A browser-based product with no installation still has to receive study context and return a signed report to the right place, and "integration available" on a website is not an answer.

Templates come second. Your existing templates and macros represent years of accumulated group convention, and they are the single most underestimated cost in any reporting migration. Ask what survives, in what format, and who does the conversion work. We wrote about who actually owns your report templates because the answer surprises people at exactly the wrong moment.

Security review comes third. Any cloud reporting platform will need a HIPAA and vendor risk assessment before protected health information reaches it, and for a smaller US practice without a dedicated privacy function that review is often the real bottleneck rather than the software. Teams that have to keep proving controls to auditors on a recurring basis usually end up wanting a structured way to track obligations and evidence rather than rebuilding the same questionnaire every renewal. Ask for the vendor's current attestations early, because procurement will ask you for them later.

Commercial terms come last and matter longest. Get the fully loaded first-year total including integration and training, then get the year-three renewal in writing before the product is embedded in a workflow you cannot easily unplug. The contract terms worth negotiating covers the uplift clause that quietly decides that number.

Is RADPAIR worth evaluating?

If you are being forced into a reporting decision by the PowerScribe 360 deadline and you are open to leaving the installed workstation behind, yes, and the published trial makes it cheap to find out. If you read across several systems, the browser-based model is a real advantage rather than a marketing line. If your integration is deep, your template library is large and your group is risk-averse about vendor age, PowerScribe One and Fluency for Imaging are the more conservative names on the same shortlist, and the PowerScribe switching guide lays out that decision properly.

What none of these products change is what happens before the dictation starts. The report is only fast if the right study is in front of you at the right time, and the reading itself is still yours. Radiological.ai flags suspected findings, prioritizes the worklist and drafts the structured report across X-ray, CT and MRI in one pane, and every flag is a suggestion the radiologist reviews, edits and signs.