Cercle
Cercle builds a data foundation for fertility care and research from records that sit in separate systems. Its Ingestion Engine extracts, anonymizes and harmonizes information from EMRs, CRMs, laboratory systems, images and video. The Biomedical Graph links patients, cycles, protocols and outcomes over time, and an Intelligence Layer adds predictions, simulations, dashboards and AI agents that answer plain language questions about patient cohorts. The homepage counts 54,568 cohorts defined on the platform.
Provider teams use the Fertility AI and Predictor tools to discuss treatment options and expected outcomes with patients, plan care and track operational measures such as conversion. Research and life sciences teams explore longitudinal patterns and treatment effects. Clinicians can edit agent outputs and send corrections back.
Customers and partners include US Fertility, a network supporting more than 100 clinics, Boston IVF, Ivy Fertility, the European distributor Eurofins Genoma and the fertility EMR MedITEX. Based in San Francisco and led by CEO Juan Carlos Riveiro, the company came out of stealth in November 2023 with backing from Sheryl Sandberg and raised a $6M seed round led by Outsiders Fund in September 2024 to expand beyond fertility. Licenses are sold after a demonstration with no public rate card, and the terms exclude diagnosis and treatment.
Capability Axes
An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read
Custom trained models, computer vision, graph based machine learning and language model interfaces turn clinical source material into cohorts and predictions. The Intelligence Layer depends on those learned components for plain language questions, patient comparisons, outcome predictions and decision support, and the agents answer cohort questions without a query language.
Ingestion, harmonization, the connected graph, dashboards and operational reporting carry value without the models, since a fertility network gains a shared view across records that sat in separate EMRs. Removing the learned components would take away the analysis and conversational tools while leaving a working data platform underneath them.
Clinicians can edit agent outputs and send corrections, output validation with clinician feedback loops is built into the tools, and traceability links each insight to the underlying data. The tools support conversations about fertility options and expected outcomes, with the clinician and patient keeping the treatment decision.
The 2024 seed announcement said the graph works without hallucinations, a claim the review mechanism does not itself demonstrate. No confidence threshold suppresses an answer, no rule forces escalation for unusual cases and no recovery process follows an incorrect prediction. Editable output gives a concrete point for human intervention without requiring approval before every clinical use.
The technical approach combines graph based machine learning, computer vision, custom trained models and language model interaction. An ingestion stage structures multimodal records, the Biomedical Graph connects events and outcomes, and the Intelligence Layer runs simulations and agent queries, with the 2024 seed round funding new LLM capabilities for that layer.
That explains what each component does without naming a production model, its version or its update history. Billions of data points describe scale rather than a reproducible training set or evaluation split, and the custom model label does not say which parts were trained in house, which were adapted, or how a customer learns that a model behind a prediction changed.
The Biomedical Graph sits between source records and the language model interfaces, with anonymization meant to remove identifying information before downstream analysis. Custom trained models and graph based learning are part of the design, but the language model provider goes unnamed. Partner clinics and labs such as US Fertility and Eurofins Genoma contribute anonymized data back into the platform.
The privacy policy lets service providers and related companies process information under data processing agreements without listing who handles inference or hosting. What each supplier receives, how long it keeps the anonymous records and whether it can reuse them therefore stay open, even with identifiers removed before the graph.
US Fertility, Boston IVF, Eurofins Genoma and Ivy Fertility are named users. The June 2025 Ivy announcement describes a network spanning four EMR systems and more than 50 physicians and lists intended improvements rather than measured changes in IVF success or conversion. US Fertility physician and Cercle advisor Eduardo Hariton describes a predictor built on eight years of network data while noting limited outcome data after its introduction.
Kate Devine, US Fertility's medical director and chief research officer, credits the platform with strengthening research and internal quality assurance. None of these accounts tests the current models, and cohort counts such as 54,568 do not supply prediction accuracy, calibration, a comparison group or patient benefit.
Patient information is meant to be anonymized before it enters the Biomedical Graph. The service privacy policy expects identifiable uploads anyway and commits to anonymizing them as soon as possible rather than keeping patient personal data. Once anonymized, information may improve the software's algorithms and support statistics and research, so training on anonymized data is permitted rather than ruled out.
The time before anonymization, retention of the anonymous records and practices at downstream model services are not stated. Editable output and clinician corrections give review controls, but no defined response to a safety event explains how affected predictions are traced and fixed.
The platform names Cercle a HIPAA business associate that ingests PHI, and its privacy policy reaches the Fertility AI Tool and Predictor AI Tool. When identifiable patient information arrives, the company acts as a processor for the healthcare center and anonymizes it under their agreement, which qualifies the wider marketing line that personal health information is never handled.
No business associate agreement or subcontractor terms can be read before contracting. How far the agreement reaches across the newer agents, the ingestion services and the research applications is therefore less clear than the business associate role itself, which is stated plainly for the two provider tools.
The platform describes encryption in storage and transit, anonymization and access restricted to licensed accounts. The privacy terms add technical, administrative and physical safeguards and place responsibility for account credentials on the user.
The trust section shows customer and partner logos, including fertility networks and MedITEX, without naming a security certification, auditor, report period or attestation, and the homepage lists no SOC 2 or ISO credential. A customer relationship does not set the security scope of the platform. The handling commitments are specific, and independent assurance of the company's own operating controls is what is missing.
The Biomedical Graph's stated position is that it has not been evaluated by the FDA and is not a medical device, and the terms exclude medical advice, diagnosis, treatment and emergency use. Current provider applications still include personalized care planning, fertility predictions and support for treatment decisions.
Those uses put more weight on the boundary than a reporting tool would. No named statutory exclusion explains how the clinical features fall outside device rules, and no clearance attaches to them. A clinician can use a prediction in a discussion without the software prescribing treatment, but that human role leaves the gap between stated scope and marketed use in place.
Traceable outputs, editable agent responses and clinician feedback loops support transparency. Advisors from fertility medicine, genetics and related fields contribute clinical expertise, and the company's mission centers on gaps in women's health data, with menopause, women's oncology and rare diseases named as next areas.
No release criteria, fairness audit or subgroup results compare predictions across ages, diagnoses, treatment histories, clinics or demographic groups. Building a cohort does not show that predictions hold within it, and monitoring and approval duties after a clinician submits a correction are not assigned. The direction is responsible use without a defined model governance framework.
The service terms give a dissatisfied user a direct route to notify Cercle and have the complaint reviewed with technical and subject matter experts. Editable agent output and clinician corrections offer a separate way to challenge a result during use.
The complaint commitment stops short of guaranteed correction or clinical accuracy. The terms disclaim particular results, cap aggregate liability at the preceding 12 months of fees and exclude consequential losses, and indemnification runs from the user to the company. A stated vendor review process exists, while no standard remedy promises compensation or repair of decisions an error affected.
MedITEX lists Cercle among its integrations, which backs a distribution and data access partnership for fertility clinics using that EMR. The platform also ingests from the four EMR systems in the Ivy Fertility network and accepts EMR, CRM, laboratory, image and video inputs.
The ingestion route is clearer than the return path. No interface specification says which clinical fields can be written back, which permissions apply or where a two way connection runs in production, and MedITEX's general HL7 and API features do not show that every such interface belongs to the Cercle connection. The named relationship supports partial interoperability.
Licensed users reach software backed by hosted servers. The privacy policy stores personal data from EEA and UK users on servers in the European Union and data from other users in the United States, and allows transfers outside the EU under applicable transfer rules. Eurofins Genoma distributes the platform in Europe, where that EU storage applies.
The regional terms cover user information rather than the full location of anonymous clinical graphs, model inference, backups or support access. Tenant separation and a customer hosted option are not described, so the arrangement is a real regional commitment without an end to end residency guarantee.
Healthcare organizations license the service after a demonstration and an engineer led sales conversation. Providers use the data foundation and clinical tools, life sciences teams use cohort exploration and research applications, and MedITEX adds a distribution route for fertility clinics.
No standard prices, tiers, usage allowances or self service purchase exist, so neither a per user nor a per patient charge can be worked out. The access terms permit charges for services and premium functions without stating amounts. Implementation, data integration and the scope of each application are settled in the commercial discussion, with no separate fees or charging units listed.
Fertility and reproductive medicine are the core, with provider applications covering IVF patient and embryo data, outcome discussions and clinic operations. Named relationships include US Fertility, Boston IVF, Ivy Fertility and the European distributor Eurofins Genoma, and life sciences teams investigate cohorts and longitudinal treatment patterns on the same foundation.
Menopause, women's oncology, rare diseases, rheumatology and other fields appear as planned uses of the architecture rather than deployments of equal depth, and the 2024 seed round was raised to fund that move beyond fertility. Model accuracy and clinical benefit are not reported across those settings or across stages of fertility care.
Alternatives to Cercle
Other products that do the same kind of work as Cercle, and what each one does.
ThetaRho sells a clinical intelligence platform in three layers and will sell any of them on its own.
Mendel, based in San Jose and backed by Oak HC/FT and DCM, sells Hypercube, a platform that reads unstructured patient records for chart review, cohort building and clinical trial prescreening.
Washington, DC company whose AI and data platform serves the medical side of life sciences: medical affairs, scientific communications and pharmacovigilance.
Komodo Health sells real world data and AI analytics to drug makers, payers, providers and the consultants and investors who serve them.
Aetion makes the Aetion Evidence Platform, software for designing and running real-world evidence studies on health data.
Clinical search and answer platform for UK clinicians and students, from Medwise AI Ltd, registered in England and Wales and based in Leeds, having started in Cambridge.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | BAA Tier | Implementation | Source |
|---|---|---|---|---|
|
Custom quote
|
Organizational software licenses cover clinical data analysis, provider tools and life sciences research. No standard charging unit, tier or rate is listed. | Business associate status is stated. The agreement, its module coverage and any tier restriction are not posted. | Integration and implementation are scoped in the sales discussion, with no separate fee listed. | Vendor Published |
Access begins with a demonstration and sales discussion. The licensed platform combines ingestion, the Biomedical Graph and intelligence applications, and MedITEX offers a further distribution route for fertility clinics. No rate per user, patient or query is listed, and the current purchase flow has no free trial.