Quris-AI

**Product and the concrete problem it solves.** Quris-AI addresses a costly translation gap in pharmaceutical development: a drug candidate can look acceptable in conventional laboratory and animal studies yet fail once tested in humans, particularly because safety and efficacy are difficult to predict across diverse patients. The company describes its product as a Bio-AI Clinical Prediction Platform rather than a pure software model. It combines automated biological experiments on miniaturized Patients-on-a-Chip with machine-learning classification to forecast which candidate drugs may work safely in humans. The initial commercial problem is concrete rather than abstract: Quris and Merck KGaA began with the identification of liver-toxicity risk, including candidates whose risks were not identified by earlier preclinical experiments. If the platform becomes reliable enough for regulated development workflows, it could help pharmaceutical teams stop weak candidates earlier, prioritize scarce laboratory capacity, and reduce the cost and elapsed time of programs that otherwise fail after substantial investment.

**Core technology and how it works.** Quris public materials describe a closed-loop system in which known safe and unsafe drugs are tested on miniaturized human tissues, next-generation nanosensors continuously measure the biological response, and the resulting labeled data is used to train and retrain a machine-learning model. The company calls the architecture AI-Chip-on-Chip: one layer is the biological model, including organ or multi-organ tissue, and another is the sensing and computational layer that converts response signals into features and predictions. Automation is central because the value proposition depends on producing many comparable experiments rather than a few bespoke organ-chip demonstrations. Quris says the platform is designed to test thousands of drugs and eventually millions of biological experiments. The company also emphasizes stem-cell-derived tissues from hundreds of genetically diverse sources, supported by an exclusive collaboration with the New York Stem Cell Foundation, to avoid training a clinical prediction model on a single narrow biological profile. Its public patent materials identify an AI-chip-on-chip clinical prediction engine that treats tissue response as evidence for classifying a drug characteristic; the claims cover elements including tissue treatment, feature extraction, prediction, microfluidic structures, and nanosensing. These are meaningful technical building blocks, but the public record does not independently establish performance across broad clinical indications.

**Market, customers, and go-to-market.** The natural buyers are pharmaceutical and biotechnology companies that need earlier, more human-relevant evidence on safety, efficacy, and disease biology. Quris has described three complementary routes to market: fee-based evaluation of external drug candidates, pharmaceutical alliances around specific disease domains, and an internal or partnered drug pipeline that uses the platform to select candidates. The first route can create a service relationship before a customer is ready to license a platform; the second can turn validated prediction into a recurring strategic relationship; and the third gives Quris a way to capture more value if its data identifies a promising therapy. The Merck arrangement illustrates the enterprise motion. Merck first received the right to assess Quris technology against traditional in-vitro and in-vivo approaches, then the collaboration expanded after a preclinical study, with a possible five-year exclusive license in a specific disease domain and undisclosed economics. Quris was also selected for AstraZeneca BeyondBio SCALE, a program involving AstraZeneca, Accenture, AWS, Clalit, and Israeli innovation institutions intended to help Israeli digital-health and AI companies deploy into healthcare systems and global markets. Quris has not publicly disclosed recurring revenue, customer count, regulatory clearance, or the number of paying pharmaceutical users, so commercial maturity remains unproven.

**Traction, funding, and third-party validation.** Quris launched publicly in October 2021 alongside a $9 million seed round led by Judith and Kobi Richter, with participation from technology and strategic angel investors. The round expanded to $28 million in January 2022 with Welltech Ventures, iAngels, GlenRock Capital, and others, and Quris announced another $9 million in December 2022 led by SoftBank Vision Fund 2 alongside existing investors, bringing disclosed seed funding to $37 million. That capital base is substantial for an R&D-stage platform and supports the expensive combination of biology, instrumentation, data generation, and pharmaceutical business development. The strongest external validation is not the funding headline but the Merck sequence: an initial collaboration assessed toxicity prediction against established methods, and Merck later extended the work after the initial preclinical study. Quris reported that the study focused on drug-induced liver injury and that its platform identified the relevant toxicity signal; this remains a company-reported collaboration result rather than a peer-reviewed clinical validation. In October 2024, Quris announced the acquisition of Nortis, recently known as Numa Biosciences, and said it would integrate Nortis Kidney-on-Chip technology that had been vetted by the National Center for Advancing Translational Sciences at the U.S. National Institutes of Health. Fast Company named Quris to its 2023 Most Innovative Companies list. These signals show sustained ecosystem attention and partner engagement, but they do not yet prove that the platform has become a regulatory-grade standard.

**Founders and team background.** Quris is led by Isaac Bentwich, MD, Founder and CEO, and Yossi Haran, Co-Founder and CTO. The public company profile describes the broader group as spanning machine learning, statistics, biology, software, genomics, engineering, and medical-device development. Bentwich is presented as a genomics-AI pioneer, while Haran is the technical co-founder associated with the chip-on-chip architecture and patent filings. The company has surrounded the operating team with unusually senior scientific and commercial advisers: its public board and advisory page names Nobel Laureate Aaron Ciechanover as chair of the Scientific Advisory Board, Moderna co-founder Robert Langer as an adviser, former Pfizer CEO Henry McKinnell as chair of the Business Advisory Board, former Biogen CEO Michel Vounatsos as a board member, and Medinol and Orbotech pioneer Kobi Richter as a board member. This bench is useful for pharmaceutical credibility, clinical translation, and access to regulated-industry decision makers. It also creates a diligence distinction: prominent advisers and investors can open doors, but they do not substitute for a large, experienced team that has taken a bioinstrumentation platform through validation, quality systems, regulatory engagement, and repeated commercial deployments. Quris does not publish a current employee count, so the scale of the operating organization is Unknown.

**Competitive dynamics.** Quris competes against several different categories rather than one direct rival. Emulate, MIMETAS, Hesperos, InSphero, and CN Bio provide organ-on-chip, tissue-model, or microphysiology platforms that compete for the same pharmaceutical preclinical budgets, although their biological models, sensing, and software architectures differ. Recursion, Schrödinger, and other computational drug-discovery companies compete for AI-enabled decision-making in the same R&D workflow, often using large biological or chemical datasets rather than Quris proprietary patient-on-chip response data. Traditional animal studies, two-dimensional cell assays, and established in-vivo or in-vitro toxicology remain the most powerful incumbent substitute because regulators and pharma scientists understand their limitations and workflows. Quris potential edge is integration: it links human-relevant tissue, high-throughput automation, continuous nanosensing, and a self-training classifier in a single data flywheel. The acquisition of Nortis can broaden the organ coverage, but it also adds integration and quality-system complexity. The central competitive question is whether Quris can demonstrate reproducible predictive lift over existing organ-chip systems, pure AI models, and standard toxicology at a cost and throughput that fits real pharmaceutical stage gates. Without comparative data, the edge is plausible and well protected by patents, but not yet established as a durable moat.

**Defense, security, and resilience dual-use relevance.** Quris has no disclosed defense customer, military contract, classified program, or fielded security capability, and its public go-to-market is pharmaceutical. Its strategic relevance is instead a credible resilience and biosecurity adjacency. A platform that can identify toxicity and efficacy signals from human-relevant tissue could, if validated for the relevant biology, help accelerate medical-countermeasure development, vaccine or antiviral prioritization, and preparedness programs that need to evaluate many candidates under time pressure. Faster and more reliable preclinical screening also reduces dependence on a narrow set of animal models and centralized laboratory capacity, which matters for health-system resilience during outbreaks or supply disruptions. The same technology could support government or defense medical research organizations in triaging compounds for infectious disease, chemical exposure, or other high-consequence health scenarios, but those are transfer opportunities rather than current company capabilities. The strategic case is strengthened by the combination of Israeli engineering, AI, genomics, and medical-device expertise and by Quris relationships with major pharmaceutical and research institutions. It is capped by the absence of public government deployments, the need for indication-specific validation, biological and data-governance constraints, and the fact that any defense adoption would require quality, regulatory, procurement, and biosafety evidence not visible in current sources.

**Growth stage, trajectory, and key diligence risks.** Quris is best classified as early: it has raised meaningful seed capital, built a technically differentiated platform, expanded a major-pharma collaboration, and acquired additional organ-chip capability, but public evidence still centers on R&D, preclinical validation, and partnership development rather than recurring revenue or approved clinical use. The trajectory to monitor is whether the company converts its integrated data flywheel into repeatable, externally reproduced prediction performance and a product that can sit inside pharmaceutical decision gates. Key diligence points are: (1) whether the tissue models reproduce clinically meaningful biology across diverse populations and disease areas; (2) whether the machine-learning outputs generalize beyond the known safe and unsafe training set without hidden label or batch effects; (3) whether Merck or another pharma customer has moved from evaluation to paid, repeatable production use; (4) whether the Nortis asset is integrated without losing assay reproducibility, personnel continuity, or intellectual-property clarity; (5) whether regulators accept Quris evidence as decision-support alongside, rather than merely adjacent to, conventional studies; (6) whether the patent portfolio protects the full integrated workflow and survives freedom-to-operate review; and (7) whether the company can finance specialized hardware, biology, quality systems, and long enterprise sales cycles. Quris is therefore a high-potential but high-risk strategic monitor: the upside is a new human-relevant safety infrastructure layer, while the failure mode is a compelling demonstration that cannot cross the validation and procurement chasm.