Nucleai

Computer vision on tissue is the product and the company argues it does something humans cannot. Its stated case is that manual pathology image analysis is subjective, inconsistent, and highly variable because it rests on pathologist consensus, while its platform characterizes cell distribution and spatial relationships at a resolution no human reader can reach.

The multimodal spatial operating system integrates high plex spatial proteomics, histopathology, and clinical data, and a separate deep learning model automates normalization of high plex imaging, a step previously done by hand.

Output feeds expert interpretation rather than replacing it. In the research setting a spatial proteomics agent generates reports from spatial datasets for translational teams to act on, and in the clinical setting the company describes its technology as used by pathologists for clinical trial patient selection. That last framing matters: a pathologist remains the decision maker even where the algorithm is embedded in trial enrollment criteria. When an algorithm determines trial eligibility, the oversight question shifts from clinical safety toward selection validity.

The platform is described with real specificity about inputs and processing: it ingests H and E, immunohistochemistry, multiplex immunofluorescence, and spatial transcriptomics data, and produces quantitative spatial insights including cell architecture and neighborhood analysis. Work has been published and presented through peer reviewed and conference channels including AACR, ASCO, and SITC, with one study selected among the top abstracts from over 1,200 submissions. Model architecture and performance figures are not published.

Nucleai's own models on its own service, with its service providers described only by category. The company builds its tissue analysis models itself and delivers results as expert-led analysis for biopharma sponsors, so no outside model provider sits behind a finding. The user terms (31 March 2026) say Nucleai may use users' input and content to develop and improve the solution, including through machine learning, unless a customer services agreement says otherwise. The website privacy policy (12 April 2026) lists service provider categories (hosting and co-location, content delivery, data security, analytics, remote access and others) without naming them, and refers to a separate Nucleai Platform Privacy Policy that is not linked from the site. The technology is also embedded in partners' digital pathology software. The platform links tissue images with genomic data and clinical outcomes, and nothing published names where that linkage runs or which parties receive the combined data.

A structurally short chain at the model layer, with the hosts and partners unnamed.

Evidence is substantial on the research side and early on the clinical side. Named pharmaceutical partnerships apply the platform to clinical stage oncology assets, academic collaborations span multiple universities on colorectal and lung cancer biomarkers, and clinical trial data has been presented at major oncology meetings, including analysis of biopsies from a named melanoma trial identifying immune cell interactions linked to outcomes.

A health system has explored the technology as a treatment decision tool in real world care. What does not yet exist publicly is prospective validation that spatial biomarkers derived by the platform improve treatment selection outcomes.

Training use is stated for users, and the document that would cover platform data is named but not published on the site. The user terms (31 March 2026) say Nucleai may use users' input and content to further develop, improve and enhance the solution, including through machine learning, subject to any customer services agreement. The website privacy policy (12 April 2026) covers website visitors and business contacts, allows creating anonymized or pseudonymized data for Nucleai and its partners to improve their services, and refers readers to a Nucleai Platform Privacy Policy for platform data, which is not linked or reachable from the site. The platform processes patient biopsy images linked to clinical outcomes and, the company has said, genomic data and medical records. No retention period, de-identification practice for tissue data or safety engineering for a wrong output is published.

General assurances, with training use stated for users and retention and de identification for patient data unanswered.

No HIPAA position is stated; a platform privacy policy is named but not published on the site. The website privacy policy (12 April 2026) is issued by Nucleai US, Inc. and its affiliates, covers website visitors and business contacts, and points to a separate Nucleai Platform Privacy Policy for platform data that the site does not link. The user terms say customer use is governed by services agreements, which are not published. The primary buyer is a biopharma sponsor rather than a covered entity, and trial data usually arrives coded, but the company also describes work with health systems and academic centers, which would bring HIPAA into scope for US data. No business associate status or agreement appears.

A named platform privacy document that cannot be read, with no status stated.

Security is asserted, and nothing independent stands behind it. The website privacy policy (12 April 2026) says Nucleai and its service providers implement systems, applications and procedures that provide sound industry-standard security, without naming any. The footer links the privacy policy and user terms only, no security or trust page exists on the company's domain, and no ISO 27001 certificate, SOC 2 report, penetration testing statement or vulnerability reporting route is published. The company receives patient biopsy images, including multiplex immunofluorescence and spatial transcriptomics, from biopharma trials and links them with clinical outcomes, so custody of identifiable-derived tissue data moves to the company, and its UK academic collaborations bring European data protection obligations into scope.

Controls asserted with nothing independent behind them.

No device claim is made, and the product is scoped to drug development. Nucleai presents itself as tissue intelligence for drug development, delivered as expert-led analysis for biopharma translational and biomarker teams, and its user terms state the solution is not intended to diagnose, treat, monitor, cure or prevent any disease or to replace a physician's judgment. FDA's 510(k) database lists no clearance for Nucleai. An investor quoted on the home page calls it the first spatial AI tool used by pathologists for clinical trial patient selection connected to a drug development program; patient selection inside a trial runs under that trial's protocol, and any future companion diagnostic would need its own regulatory route.

No device claim, with the product scoped accordingly.

The company names the variability problem and has built a tool for it; the residual variability is not measured. Nucleai has stated that manual normalization of high plex imaging is inconsistent across laboratories and users, and released a deep learning model to automate it, which it says improves reproducibility and runs nearly twenty times faster than manual work. The speed gain is quantified; the reproducibility gain is asserted without a figure, and no between-site, between-scanner or between-staining-platform results are published, nor the conditions under which normalization fails. No governance framework or evaluation by patient group is published. Published biomarker work rests on a retrospective discovery cohort of about 90 patients in one indication, validated on about 43, which is small for a product used in trial patient selection.

Variability acknowledged with no evaluation by group or site published.

Findings are traceable to their source images, and the terms disclaim the rest. The home page says every finding can be traced back to the original image it came from, and analysis is delivered as expert-led work under a Quality by Design approach, so a sponsor's team can check a finding against the tissue. The user terms (31 March 2026) disclaim warranties of accuracy and reliability, state that the solution is not intended to diagnose, treat or replace a physician's judgment, and exclude liability to users who have made no payment; customer services agreements, which are not published, may say more. No error rate with a method and denominator is published for any spatial signature.

Source traceability and expert review, with nothing standing behind a finding.

Integration is with the research pathology stack rather than the chart, and no system is documented. The company has partnered with a digital pathology platform vendor to embed its biomarker technology in that vendor's software, works with spatial biology instrument platforms, and links tissue findings with clinical outcomes data supplied by sponsors. No EHR, HL7 or FHIR interface, LIS connection or integration documentation is published, which fits a service delivered to biopharma teams rather than to clinical systems.

Integration described through partnerships with nothing to verify.

Described as a cloud based end to end analytics platform processing multiplex immunofluorescence, H and E, immunohistochemistry, and spatial transcriptomics data. No tenancy, hosting, or residency terms are published, which matters for a company headquartered in Israel serving pharmaceutical partners globally.

No published pricing. The commercial model appears to combine pharmaceutical partnership agreements with platform access, and neither structure nor economics are disclosed.

Drug development settings, stated plainly, with research validation behind part. Nucleai works for biopharma sponsors wherever tissue biology is central to a program decision, across oncology, immunology and metabolic disease and drug modalities such as antibody drug conjugates and multispecifics, supporting translational medicine, clinical biomarker and translational pathology teams; the home page reports work across more than 12 clinical trials and more than 30 indications, and a collaboration with Gilead analyzing H&E and IHC slides across several clinical studies of its ADC programs. Published work includes a retrospective discovery cohort of about 90 patients in one indication with validation on about 43. Clinical diagnostic use is not the current scope, and the user terms state the solution is not intended to diagnose or treat.

Coverage named with validation behind part of it.