Harrison.ai vs Behold.ai (2026)
Two chest radiograph companies separated by whether a human still reads the film. Annalise reads comprehensively across a large findings set with regulatory footprints in several markets and greater deployment maturity, and the radiologist decides every case. Behold auto reports films it classifies as high confidence normal, taking them off the worklist entirely, and publishes the peer reviewed performance numbers that autonomy demands. For a department whose backlog is normal films, Behold is doing something Annalise does not attempt and the saving is structural rather than incremental. The condition is that its false negative rate at the operating threshold has to be published, revalidated locally and monitored, because on an auto reported study nobody is ever going to catch the miss.
- Comprehensive chest radiograph coverage across a large findings set, with regulatory footprints in multiple markets and more than a third of NHS England chest X-rays.
- The reader retains the decision on every study, which is the conventional and lower risk configuration.
- Deployment maturity is greater, which matters for a modality where the failure is a subtle finding missed at scale.
- It removes studies from the worklist entirely by auto reporting films classified as high confidence normal, which is genuine autonomy inside a bounded task.
- The performance numbers that autonomy requires are published in peer reviewed venues, and the confidence threshold is itself the governance mechanism.
- For a service drowning in normal chest films, removing them from the queue is a different economic proposition from reading them faster.
This comparison is published by AI Health Index, an independent research platform that compares healthcare AI vendors objectively. Harrison.ai and Behold.ai are each graded against the same capability taxonomy, from each vendor's own public materials and the regulatory record, under the AI Health Index verification standard. No vendor pays for placement, and no vendor has reviewed this page. How this evidence is graded
Plain facts
| Fact | Harrison.ai | Behold.ai |
|---|---|---|
| Primary category | Radiology & Imaging AI | Radiology & Imaging AI |
| Founded | 2019 | 2015 |
| Headquarters | Sydney, Australia | London, United Kingdom |
| Website | harrison.ai | behold.ai |
Side by Side
Select any grade to read the note behind it.
Each vendor in one paragraph
Written to be quoted whole. Each paragraph opens the AI Health Index profile of the vendor: what the product is, who buys it and what it does.
Harrison.ai, headquartered in Sydney, sells radiology AI that until 2025 was sold under the Annalise.ai name. Harrison.ai combined with Annalise.ai in February 2025 and renamed the Annalise products in October 2025, and the annalise.ai domain now forwards to harrison.ai.
Source: AI Health Index, August 2026
UK company whose red dot platform sits at the autonomous end of imaging AI: it reports chest X-rays it classifies as High Confidence Normal autonomously, removing them from the radiologist worklist (about 15 percent of studies, according to the company), triages suspected lung cancer for urgent reporting, and has a companion CT head product. Work published in Clinical Radiology reported a 0.33 percent error rate on the auto reported studies and a 60 percent reduction in missed lung cancers when triage ran alongside consultant radiologists.
Source: AI Health Index, August 2026
Questions buyers ask
How do Harrison.ai and Behold.ai work with the EHR?
For Harrison.ai, the 510(k) summary for K222268 describes the software receiving DICOM chest X-ray studies from image and order management systems such as a PACS or RIS, running in parallel with the standard workflow, and returning a case level result to the PACS or RIS for worklist prioritization.
With Behold.ai, the how it works page states that after analysis it sends an electronic notification back to the trust RIS, or to PACS where reporting is PACS driven, marking each examination as normal or abnormal and prioritizing it in the existing worklist; for auto reported studies that notification is the report. From the AI Health Index, drawn from each vendor's own published materials and the regulatory record, verified August 4, 2026. No vendor pays for placement.
How are Harrison.ai and Behold.ai priced and sold?
In Harrison.ai's case, the products are sold to healthcare organizations (public hospitals, private hospitals and private radiology providers) by contacting the company, and the site states they are not available for purchase by the general public.
For Behold.ai, the original company entered administration in January 2025 after failing to win NHS AI Diagnostic Fund contracts, NICE withdrew the product from its recommendations the following month, and a successor acquired the assets and relaunched it. From the AI Health Index, drawn from each vendor's own published materials and the regulatory record, verified August 4, 2026. No vendor pays for placement.
How do Harrison.ai and Behold.ai handle patient data?
With Harrison.ai, the company's privacy notice (last updated 20 November 2025), which covers its products and services as well as its websites, states that it does not use any content uploaded into its services, including demos, product trials and evaluations, for model training.
In Behold.ai's case, the how it works page states that images reach the platform over an encrypted VPN and that all patient examination data resides in fully NHS accredited data centers. From the AI Health Index, drawn from each vendor's own published materials and the regulatory record, verified August 4, 2026. No vendor pays for placement.
Which settings and specialties do Harrison.ai and Behold.ai cover?
For Harrison.ai, the company sells chest X-ray (up to 124 findings), non contrast head CT (up to 130 findings) and chest CT (up to 167 radiological features) to public hospitals, private hospitals and private radiology providers, with solution pages for emergency and acute care and for oncology management, and states clearance for clinical use in more than 40 countries. The US triage clearances carry defined, validated populations.
With Behold.ai, the settings on record are NHS trust radiology departments. From the AI Health Index, drawn from each vendor's own published materials and the regulatory record, verified August 4, 2026. No vendor pays for placement.
Related comparisons
Other published head to head assessments involving these vendors or their closest peers. The full set for this category is on the Clinical Decision Support page.
The autonomy difference sets the evidence bar: an auto reporting normal classifier needs its false negative rate at the operating threshold published and locally revalidated, because a missed finding on an auto reported film is never seen by anyone. Behold's regulatory standing is multi jurisdictional and unsettled and its corporate history complicates the identity of the contracting entity, both of which should be established before deployment. Neither publishes a security attestation or pricing, and findings counts are not comparable between them.