Rad AI

What is Rad AI?

Rad AI is a radiology-focused AI platform that drafts report impressions, provides a cloud reporting system and tracks follow-up on incidental findings. It sells to health systems and radiology practices rather than to individual users.

The platform has three products. Rad AI Reporting is a cloud-native reporting system with speech recognition tuned for radiology. Rad AI Impressions writes the impression section from dictated findings in each radiologist’s own style, and Rad AI Continuity reads finished reports to find recommended follow-up. The radiologist reviews and signs every report, so the software supports clinical work and does not give medical advice or make diagnoses.

The company was founded in 2018 and is based in the Bay Area. Its about page names radiologist Jeff Chang, MD, as co-founder and president and Doktor Gurson as co-founder and chief executive. The site claims more than 11,000 radiologist users, over 200 health organisations and more than $140 million raised from investors including Khosla Ventures and Transformation Capital. In June 2026, AuntMinnie reported that Yale New Haven Health would deploy Rad AI’s reporting system across its imaging network.

Key Features

Personalised impressions sit at the centre of Rad AI, because the model learns each radiologist’s phrasing from past reports and drafts the summary section for them.

• Personalised impressions: Rad AI Impressions writes the impression from dictated findings, matching the radiologist’s usual wording and clinical preferences. The official site says it works inside PowerScribe and Fluency, so radiologists keep their current dictation system and simply approve or edit the draft.
• Cloud reporting: Rad AI Reporting is a cloud-native reporting system with continuous speech recognition tuned for radiology. Radiologists can dictate findings in any order, and the software turns free-form speech into a structured report with templates, macros and voice commands.
• Guideline recommendations: Drafted impressions can include follow-up wording based on consensus guidelines. The Impressions page names Fleischner, incidental thyroid nodule and abdominal aortic aneurysm guidance as examples of recommendations the system inserts for the radiologist to review.
• Pre-sign-off quality checks: A real-time checker flags inconsistencies in a report before the radiologist signs it. Catching problems at this stage can reduce the number of corrections and addenda needed after a report has already reached the referring clinician.
• Carry-forward for follow-ups: For follow-up exams, unchanged findings from the prior report can be pre-generated automatically. Rad AI also lists automated formulas and calculations, which together cut repetitive dictation on patients who return for repeat imaging.
• Follow-up tracking: Rad AI Continuity classifies follow-up recommendations across more than 50 categories and 250 subcategories of findings. It sets due dates and sends reminders by EHR message, print mail or fax, keeping an audit trail of every contact.

Pros

Time saved per report is the strongest case for Rad AI, with the company citing up to 50 per cent faster reporting and far fewer dictated words.

• Faster reporting: Rad AI says its reporting system can make report completion up to 50 per cent faster. An ARA Health case study on its site reports a 20 per cent median time reduction, with top-quartile users reaching 40 per cent.
• Less dictation: Impressions claims an 80 to 90 per cent cut in words dictated for the impression section. Radiologists reading high volumes gain the most, because each saved sentence repeats across a full worklist.
• Works with existing dictation: Impressions runs alongside PowerScribe and Fluency, so practices do not need to replace their reporting system first. This lowers switching cost and lets a group trial AI impressions before any larger platform change.
• Closes the follow-up loop: Continuity case studies cite twice the patient follow-up adherence and completion rates above 70 per cent. Health systems gain a tracked process for incidental findings that could otherwise be lost between departments.
• Radiology-only focus: A radiologist co-founded the company, and the homepage lists more than 15 radiologist employees and advisers. Its models are said to be trained on nearly a billion radiology reports, which suits specialist terminology better than general dictation tools.

Cons

Rad AI publishes no price list, so practices cannot compare its cost with other reporting tools without first going through a sales process.

• Quote-only pricing: No prices appear on the official site, and directories such as Futurepedia simply list it as contact for pricing. Budget holders must request a demo and a quote before they can judge value against current dictation costs.
• Not for individuals: There is no self-serve sign-up, free plan or public trial for a single radiologist. The products are deployed through organisations, which means contracts, IT involvement and training before anyone can use them.
• Radiology only: The platform covers radiology reporting and follow-up rather than wider clinical documentation. Health systems that want one AI assistant for every specialty will still need a separate tool for non-imaging departments.
• Review is still essential: Generated impressions and guideline text are drafts that can contain errors or misplaced recommendations. Radiologists must check each one before signing, and real time savings depend on how quickly they can edit.
• Limited public security detail: Encryption standards, hosting regions and data use for model training are not described on the security page. Formal policies are available on request, so governance teams must ask for them during procurement.

Who is Using Rad AI?

Radiology practices and US health systems make up Rad AI’s customer base, and the company says it works with over 40 per cent of US health systems.

• Radiology groups: Large practices such as LucidHealth and Radiology Associates of North Texas are named as customers. They use the impression and reporting tools to raise reading throughput without longer working days.
• Health systems: Organisations including Emory, Trinity Health and AdventHealth appear on the homepage. They use the platform to bring consistent reporting and follow-up processes to imaging across many sites.
• Follow-up coordinators: Quality and navigation teams at systems such as Guthrie and Cone Health use Continuity. It helps them track incidental findings and contact patients and referring providers on time.
• Academic medical centres: Yale New Haven Health announced in June 2026 that it would roll out Rad AI’s reporting system. Its imaging network handles more than 700,000 radiology exams a year.

Pricing

Rad AI sells on quote-based enterprise contracts and lists no prices, plans or free tier on its website (official site, October 2026).

Futurepedia also lists Rad AI as contact for pricing, and none of the directories we read gave a figure.

Disclaimer: cost will depend on exam volume, the products you combine and integration work, so ask for a written quote that includes implementation and training.

What Makes Rad AI Unique?

Personalisation to the individual radiologist is the clearest difference, because Rad AI Impressions writes in each user’s own phrasing rather than a single house style. The company says its current model is its eighth generation, built on nearly a billion radiology reports, and it collects feedback from users to keep improving drafts.

Breadth is the second difference. Many radiology AI products handle one step, such as dictation or image triage, while Rad AI covers dictation, impression drafting and follow-up tracking from one vendor. A practice can start with Impressions inside its existing system and later move to the full reporting platform.

Compatibilities and Integrations

Standards-based connections are how Rad AI joins hospital systems, using HL7, FHIR and DICOM-SR to link with PACS, RIS and EHR software.

• Dictation systems: Impressions works inside PowerScribe and Fluency, according to the official product page.
• EHR and RIS: Continuity offers bi-directional integration with Epic and Cerner through HL7 and FHIR, with status updates triggered by HL7 events.
• Deployment: Reporting is cloud-native software, with only a lightweight HL7 edge agent installed on site and updates applied automatically.
• Sign-in: SAML and OpenID Connect are supported for single sign-on with hospital identity systems.
• Security: Rad AI states SOC 2 Type II certification, HIPAA compliance, a radiology-specific de-identification pipeline and yearly third-party audit cycles.

Rad AI Tutorials

Rollout starts with your organisation, since Rad AI is deployed after a demo, contract and integration project rather than downloaded by a single user.

1. Request a demo through the Rad AI website and agree which products your group needs.
2. Work with Rad AI and your IT team to connect PACS, RIS, EHR and dictation systems.
3. Complete training and let Impressions learn from your historical reports.
4. Dictate findings as usual and review the drafted impression before signing.
5. Check any inserted guideline recommendations against the patient’s clinical context.
6. Use the feedback function and adoption reports to refine drafts over time.

How We Rated It

Strong time savings and deep radiology focus give Rad AI 4.0 out of 5 overall, while quote-only pricing pulls its cost score down.

Summary

Rad AI suits radiology groups and health systems that want faster reports and tracked follow-up from one radiology-focused vendor. The main limitation is access, since there is no public pricing and no option for an individual radiologist. Every impression remains a draft, so the tool supports radiologists and is not a source of medical advice. We rate it well for workflow depth and integration but lower for cost transparency. If you are assessing it, pilot Impressions inside your current dictation system first and measure turnaround times before and after.