> ## Content Index > Fetch the complete content index at: https://www.theaidirectory.com.au/llms.txt > Use this file to discover other available public pages before exploring further. # Annalise.ai (Harrison.ai) Review: Australia's Most Advanced Radiology AI - URL: https://www.theaidirectory.com.au/review-annalise-australia-2026/ - Published: 2026-06-25T23:00:00.000Z - Updated: 2026-10-09T06:27:13.000Z - Description: Annalise.ai is a TGA-cleared AI radiology platform built in Sydney, capable of detecting 124+ findings on chest X-rays and 130+ on head CTs. A genuine world-class tool — but it's for radiologists and hospitals, not patients or GPs. - Author: Adam - Tags: Healthcare & Medical AI, AI Reviews, #review **20**/25 ★★★★☆ Very Good Annalise.ai (Harrison.ai) · Harrison.ai (Harrison-AI Medical Pty Ltd), Sydney NSW Verified 26 Jun 2026 Value for Money**4**/5 Ease of Use**4**/5 Feature Depth**5**/5 Australian Fit**4**/5 Reliability & Support**3**/5 Pricing Pricing on application — see the pricing table **The gate:** Five stars need 21+ and Australian Fit of at least 4/5. [Skip to the verdict ](#verdict) > **Verdict (TL;DR):** Annalise.ai is one of the world's most capable radiology AI systems, built in Sydney and TGA-cleared for clinical use in Australia. It detects more findings on chest X-rays than virtually any competitor, integrates cleanly into existing hospital PACS and RIS workflows, and is already live at over 1,000 sites globally — including across the Australian public health system. This is not a consumer tool. It is enterprise clinical software sold direct to hospitals, radiology networks, and health systems. Pricing is not public. If you are a radiologist, health system IT lead, or medical administrator evaluating AI diagnostic support, Annalise is worth serious consideration. If you are a patient, a GP, or a small practice owner, this review is background reading only. > > **Score: 20/25 — ⭐⭐⭐⭐ Very Good** > > *This tool does not replace the clinical judgement of a qualified radiologist. AI-generated findings must always be reviewed and verified by a licensed medical practitioner. Last reviewed: 26 June 2026 by Adam.* --- ## What Annalise.ai is Annalise.ai is the flagship product of Harrison.ai, a Sydney-based medical AI company founded in 2018 by Dr Aengus Tran and Dimitry Tran. The company originally operated as two separate entities — Harrison.ai (the parent) and Annalise.ai (the radiology product) — before merging into a single brand. The product line now sits under the Harrison.ai name, though "Annalise" remains the market-facing label for the radiology AI tools. The core offering is an AI decision-support system for radiology. It analyses DICOM medical imaging — specifically chest X-rays and CT scans — and flags clinically significant findings for radiologist review. It does not replace the radiologist. It works alongside them, triaging the worklist, highlighting critical findings, and providing a structured second opinion that speeds up reporting and catches what might otherwise be missed. Harrison.ai raised US$112 million in a Series C round in February 2025, valuing the company above US$1 billion. It has tripled its contracted annual recurring revenue for three consecutive years. These are not startup vanity metrics — they indicate a clinical AI product that hospitals are buying, deploying, and renewing. Harrison.ai is headquartered in Sydney. It trained its AI substantially on Australian imaging data collected in collaboration with I-MED Radiology Network, Australia's largest radiology provider. --- ## What it costs in Australia Annalise.ai is enterprise software. There is no public pricing. Hospitals, radiology networks, and health systems negotiate contracts directly with Harrison.ai's commercial team. The pricing structure is annual-recurring-revenue (ARR) software licensing — standard for clinical AI platforms — but specific per-scan, per-site, or per-seat figures are not disclosed publicly. (no public pricing page exists; confirmed enterprise sales-led model) This means Annalise.ai is not accessible to individual GPs, private practices, or patients. It requires a health system-level procurement process. Contact Harrison.ai at harrison.ai to arrange a demonstration and commercial discussion. --- ## How it works in practice Annalise.ai integrates into a hospital's existing Picture Archiving and Communication System (PACS) and Radiology Information System (RIS). When a scan arrives, Annalise processes it alongside normal workflow. It produces a structured finding report — highlighting up to 124 potential findings on chest X-rays, up to 130 on non-contrast head CTs, and up to 167 radiological features on chest CTs — and returns these to the radiologist's worklist. The worklist prioritisation feature is clinically significant. Annalise can flag critical cases — suspected pneumothorax, mass lesions, signs of stroke — for immediate attention, while marking genuinely unremarkable scans accordingly. In a busy public hospital emergency department, this kind of triage has direct patient-safety value. The AI backend can be hosted on Annalise's cloud infrastructure, or deployed to meet institutional data requirements. Integration is managed by Harrison.ai's professional services team, who configure the connection between hospital infrastructure and the Annalise backend. Annalise is currently operational at more than 1,000 healthcare facilities globally — including all public emergency departments in Hong Kong and more than 40 NHS Trusts in the UK. In Australia, the company claims availability to more than 50% of Australian radiologists. --- ## What it does well **Breadth of findings.** Most radiology AI tools detect 10–20 findings. Annalise detects 124 on chest X-ray alone. That breadth matters clinically — incidental findings, subtle fractures, early-stage disease — things that get missed in a busy emergency department. **TGA clearance.** Harrison-AI Medical Pty Ltd holds two ARTG registrations: ARTG 454945 (Radiology DICOM image processing application software) and ARTG 513346 (X-ray image interpretive software). These are not trivial. TGA clearance means the tool has been assessed against Australian regulatory standards for safety and performance as a medical device. It is a hard requirement for any AI tool used in clinical diagnosis in Australia. **Australian origin and AU deployment.** Harrison.ai is a Sydney company. Its AI has been trained substantially on Australian imaging data. Annalise is not an offshore product retrofitted for the Australian market — it was built here for here. **Worklist integration.** Rather than requiring radiologists to use a separate interface, Annalise plugs into the workflow they already use. That is a fundamental design principle that competitors frequently get wrong. **International regulatory coverage.** Beyond TGA, Annalise holds EU CE MDR Class IIb certification and FDA clearance in the US. For Australian health systems with international affiliates, or procurement teams who need robust regulatory coverage across jurisdictions, this matters. --- ## What it doesn't do well **No public pricing.** The complete absence of published pricing is a real friction point for health system procurement teams doing initial due diligence. Any evaluation requires direct engagement with Harrison.ai's sales team before you know whether the product is commercially viable for your context. **Not accessible outside large institutions.** Private radiologists, regional imaging centres, and rural hospitals without dedicated IT teams face significant barriers. The integration process requires professional services engagement and enterprise-grade IT infrastructure. **Training data consent — flagged, not fully resolved.** In September 2024, reporting revealed that I-MED provided patient scan data to Annalise.ai for AI training between 2020 and 2022, without patients' explicit knowledge. The Office of the Australian Information Commissioner investigated and closed its inquiry in 2025, concluding the data had been de-identified sufficiently that it no longer constituted personal information under the Privacy Act. However, independent experts noted that Australians could reasonably expect to be asked to consent before their sensitive health data is used to train commercial AI. The legal conclusion is clear; the ethical question is not. Institutional buyers should factor this into their governance assessments. **Cloud connectivity dependency.** The standard deployment model relies on outbound internet connectivity from hospital infrastructure to Annalise's cloud backend. Institutions with strict network isolation requirements may find this constraining and should discuss on-premises or hybrid options with Harrison.ai. **AI does not replace professional assessment.** Annalise is decision support, not decision making. Every finding it surfaces must be reviewed and verified by a licensed radiologist. Clinical responsibility remains with the practitioner who signs the report. --- ## Australian fit — TGA, AHPRA, health data, and clinical safety This is where Annalise.ai genuinely earns its credentials. **TGA:** ARTG registrations 454945 and 513346 confirm TGA clearance for two distinct product categories. This is a prerequisite for clinical deployment in Australia. **Data residency:** The Annalise backend can be hosted in Australian cloud infrastructure. Data residency arrangements should be confirmed with Harrison.ai's commercial team for any specific deployment — particularly in the context of the Privacy Act 1988 (Cth) and the Australian Privacy Principles, which apply to health information collected from patients. **My Health Record:** Annalise does not appear to integrate directly with the My Health Record system. (confirm with vendor.) It operates within hospital PACS/RIS environments rather than at the consumer health record layer. **AHPRA:** Annalise is a decision-support tool for AHPRA-registered radiologists. It does not perform autonomous diagnosis. Clinical responsibility remains with the radiologist reviewing and signing the report. **Training data history:** The 2024 I-MED/Harrison.ai data-sharing matter — and the OAIC's subsequent investigation — is relevant context for any institutional buyer. The legal outcome was favourable to Harrison.ai, but the episode underlines the importance of robust data governance frameworks when deploying clinical AI. Procurement teams should request Harrison.ai's current data use policies and confirm compliance with the Health Records Acts applicable in their jurisdiction. **Australian company:** Harrison.ai is incorporated in Australia, headquartered in Sydney, and backed substantially by Australian capital. That matters for regulatory risk, support access, and procurement confidence in the healthcare sector. --- ## Alternatives For health systems evaluating clinical AI in imaging, the relevant comparison set is narrow: - **Microsoft Dragon Copilot** ([review](https://www.theaidirectory.com.au/review-microsoft-dragon-copilot-australia-2026/)) — clinical AI focused on ambient documentation and dictation rather than imaging analysis. Different use case, but part of the same hospital AI procurement conversation. - **Heidi Health** ([review](https://www.theaidirectory.com.au/review-heidi-health-australia-2026/)) — Australian AI medical scribe for GP and specialist consultations. Not imaging AI, but a useful companion for clinicians reducing admin burden. - **Nabla** ([review](https://www.theaidirectory.com.au/review-nabla-australia-2026/)) — AI clinical documentation and ambient scribe. Not a radiology tool, but relevant for the broader clinical AI category comparison. For direct radiology AI competitors — Aidoc, Rad AI, Nuance PowerScribe — none hold the same TGA-cleared product depth for the Australian market. Annalise is the local benchmark. --- ## Bottom line Annalise.ai is the standard against which other radiology AI in Australia should be measured. It is TGA-cleared, deployed at scale, and built by a Sydney company with genuine clinical credibility and regulatory rigour. The 124-finding chest X-ray detection capability is not marketing language — it reflects years of development on Australian and global imaging data. The limitations are real: enterprise-only access, opaque pricing, a legitimate (though legally resolved) data consent controversy, and cloud deployment requirements that may not suit all institutions. None of these are deal-breakers for a serious health system buyer. But they need to be on the checklist. For radiologists and health system decision-makers, this is where the bar is set in Australia. --- ## Score | Criterion | Score | Notes | | --------------------- | --------- | ------------------------------------------------------------------------------------------------------------------------- | | Value for Money | 4/5 | No public pricing — enterprise SaaS model; confirmed ARR growth suggests hospitals renew; assumed competitive given scale | | Ease of Use | 4/5 | PACS/RIS integration well-executed once deployed; initial setup requires professional services | | Feature Depth | 5/5 | 124 findings on CXR, 130 on CT head, 167 on chest CT — unmatched breadth in the AU market | | Australian Fit | 4/5 | TGA-cleared (ARTG 454945 + 513346), Sydney HQ, AU-trained data, AU data residency available; data consent history noted | | Reliability & Support | 3/5 | No public uptime SLAs; enterprise professional services; 1,000+ global deployments suggest operational stability | | **Total** | **20/25** | **⭐⭐⭐⭐ Very Good** | *Australian Fit 4/5 meets the Australian Fit gate for Very Good. Raw score of 20/25 does not reach the 21/25 threshold for Excellent.* --- ## Frequently asked questions ### Is Annalise.ai approved by the TGA in Australia? Yes. Harrison-AI Medical Pty Ltd holds two TGA ARTG registrations: ARTG 454945 (Radiology DICOM image processing application software) and ARTG 513346 (X-ray image interpretive software). These confirm regulatory clearance for clinical use in Australia. ### Can a GP or small medical practice use Annalise.ai? No. Annalise.ai is enterprise clinical software sold directly to hospitals, radiology networks, and health systems. It requires integration into hospital PACS and RIS systems and a commercial contract with Harrison.ai. It is not available to individual practitioners or small practices. ### What did the OAIC find about Annalise.ai's use of patient data? In September 2024, reports revealed that I-MED Radiology provided patient scan data to Annalise.ai for AI training between 2020 and 2022 without patients' explicit knowledge. The Office of the Australian Information Commissioner investigated and closed its inquiry in 2025, concluding the data had been de-identified sufficiently to not constitute personal information under the Privacy Act. Independent experts noted the ethical expectation of consent remains reasonable despite the legal outcome. --- ## Sources - Harrison.ai official site: [harrison.ai](https://harrison.ai/?ref=theaidirectory.com.au) — [Wayback 30 April 2026](https://web.archive.org/web/20260430180301/https://harrison.ai/) - Annalise.ai product page: [annalise.ai](https://www.annalise.ai/home?ref=theaidirectory.com.au) - TGA ARTG 454945: [tga.gov.au/resources/artg/454945](https://www.tga.gov.au/resources/artg/454945?ref=theaidirectory.com.au) - TGA ARTG 513346: [tga.gov.au/resources/artg/513346](https://www.tga.gov.au/resources/artg/513346?ref=theaidirectory.com.au) - Harrison.ai Series C — TechCrunch, 11 February 2025: [techcrunch.com](https://techcrunch.com/2025/02/11/australian-healthtech-startup-harrison-ai-scores-112m-series-c/?ref=theaidirectory.com.au) - Harrison.ai and Annalise.ai unite — Harrison.ai blog: [harrison.ai/news](https://harrison.ai/news/harrison-ai-and-annalise-ai-unite-as-one-company-to-accelerate-global-growth/?ref=theaidirectory.com.au) — [Wayback 25 January 2026](https://web.archive.org/web/20260125065812/https://harrison.ai/news/harrison-ai-and-annalise-ai-unite-as-one-company-to-accelerate-global-growth/) - OAIC statement — closure of inquiries with I-MED, Harrison.ai and Annalise.ai: [oaic.gov.au](https://www.oaic.gov.au/news/media-centre/statement-on-closure-of-preliminary-inquiries-with-i-med,-harrison.ai-and-annalise.ai?ref=theaidirectory.com.au) — [Wayback 14 February 2026](https://web.archive.org/web/20260214050326/https://www.oaic.gov.au/news/media-centre/statement-on-closure-of-preliminary-inquiries-with-i-med,-harrison.ai-and-annalise.ai) - Patient data controversy — Crikey, September 2024: [crikey.com.au](https://www.crikey.com.au/2024/09/19/patient-scan-data-train-artificial-intelligence-consent/?ref=theaidirectory.com.au) - The Conversation analysis of I-MED/Harrison.ai data sharing: [theconversation.com](https://theconversation.com/an-imaging-company-gave-its-patients-x-rays-and-ct-scans-to-an-ai-company-how-did-this-happen-245753?ref=theaidirectory.com.au) — [Wayback 3 September 2026](https://web.archive.org/web/20260903183935/https://theconversation.com/an-imaging-company-gave-its-patients-x-rays-and-ct-scans-to-an-ai-company-how-did-this-happen-245753) - Austrade case study: [austrade.gov.au](https://www.austrade.gov.au/en/news-and-analysis/news/annalise-ai-helps-improve-patient-care-worldwide?ref=theaidirectory.com.au) — [Wayback 17 February 2026](https://web.archive.org/web/20260217082336/https://www.austrade.gov.au/en/news-and-analysis/news/annalise-ai-helps-improve-patient-care-worldwide) --- --- *The AI Directory Australia is an independent editorial publication. We don't accept payment for positive reviews. Read our* [*Methodology*](https://www.theaidirectory.com.au/methodology/) *and* [*Affiliate Disclosure*](https://www.theaidirectory.com.au/affiliate-disclosure/)*.*