Incredible Health

Models drive the core modules of a marketplace that also carries human and network value. The marketplace page says matching algorithms surface best-fit candidates across more than 90 data points, including 250+ skills, specialties, licenses and geographic preferences. Lyn, sold to employers, is a voice agent that conducts phone interviews with candidates around the clock and delivers structured summaries and recordings to recruiters; Gale, offered to nurses, drafts resumes with AI and runs mock interviews with an AI voice interviewer that scores the answers. Understanding speech and generating language are functions whose working form is a trained model, so the learned component is found on the company's own pages, and the company says its AI is trained on millions of marketplace interactions and refined with each hiring cycle.

The product is not only the models: career advocates engage matched candidates and confirm interest, client partners work with each employer, and the value of the marketplace depends on the supply of nurses on it. Lyn's interviews would not exist without the model.

The oversight structure is described and the part before the recruiter is missing. The terms tell candidates that Lyn facilitates the interview and that a human recruiter reviews the results and makes all final decisions about their candidacy; recruiters receive a structured summary with the recording, and the marketplace page describes matches that career advocates engage and confirm before an employer's recruiter or Lyn screens by phone. An employer can choose to run phone screens with its own recruiters instead of Lyn.

What is not published is the step before any human looks: how the matching model decides which candidates an employer sees, whether a recruiter can view candidates it ranked low, and whether a candidate is told she was considered and not surfaced. In a ranked marketplace, a nurse who never appears in a recruiter's list has been screened out without anyone deciding to. The Lyn page's "100% applicant coverage" is a coverage claim about who gets interviewed, not a claim that the interviews are right.

The approach is described in general terms and nothing is identified. The company's AI information page says its technology is trained on millions of healthcare hiring interactions across 10,000 specialty roles, 90 candidate data points and 250 clinical skills, and that each hiring cycle refines its data models and matching algorithms; the Lyn page calls the agent healthcare-trained and voice-first. No model, model family or provider is named for matching, Lyn, Gale's resume writer or its mock interviewer, no version or release practice is published, and nothing says which of the 90 data points the matching model uses or what outcome it is trained to predict.

That last question decides what the model learns: trained on which candidates employers chose, it learns employer preferences; trained on which hires stayed, it learns something closer to fit.

No model provider is named, though the data architecture is laid out. The privacy policy and terms describe how data moves: voice interviews recorded and transcribed for the hiring team, licensure and certification records obtained from public databases, candidate data imported from employers' applicant tracking systems, and disclosure to service providers, affiliates, potential employers, advertising networks, analytics providers and social networks.

None of the service providers is named, there is no subprocessor list, and nothing says whether speech recognition, transcription or language generation for Lyn and Gale runs on the company's own models or a third party's. The trust center on Vanta may carry the provider list.

Named deployments are published with dated before and after figures, measured by the company. The case studies report that Johns Hopkins Health System cut time to hire for permanent specialty roles from 95 days to 21 days, a 78 percent reduction within two quarters, using matching and Lyn-screened interviews; that Montefiore saved $3.8 million on premium labor in the first year and sourced 33 percent of its nurse hires through the marketplace; that Providence doubled permanent placements across 51 hospitals year over year; and that at Alameda Health System 70 percent of hires were for hard-to-fill specialties and 21 percent relocated from out of state. The Lyn page reports interviews completed in 2 days rather than 15, 93 percent completion against 10 percent for manual screening and 96 percent candidate satisfaction.

The headline figures carry less method. The marketplace page says hires come in under 20 days against a US average of 86 days (the company's AI information page gives 82 days) and that its hires have 15 percent higher retention than industry averages, with no source for the averages, no cohort and no period. None of the figures is independently evaluated, and none measures whether the matched hires performed better in the role.

The privacy policy (effective 21 July 2026) states what is collected in detail and leaves the questions an AI product raises open. It says Lyn interviews are recorded, that licensure and certification records are obtained from public databases and used to customize the experience with the AI agent, that data is used for product improvement and to develop new offerings, and that aggregated, anonymized, pseudonymized or de-identified data sets are created and shared with third parties for marketing, advertising and research. The company's AI information page says its models are trained on millions of marketplace interactions and learn from every hiring cycle, which discloses that candidate activity trains the models, with no option to decline stated.

No retention period is given for profiles, recordings or transcripts, and the terms' statement that interview data is used strictly for recruitment sits beside a policy that permits advertising and research uses. Safety engineering, meaning how a wrong or harmful model output is caught and handled, is not published. No patient information is handled; the subject is a nurse's professional record, and the paying customer is the employer.

A substantive privacy policy covers the service, and no business associate status or scope position is stated. The privacy policy (effective 21 July 2026) covers the site, apps and services, including Lyn recordings, licensure records and data imported from employers' applicant tracking systems, and says nothing about protected health information. The product's subjects are clinicians rather than patients, and no patient data is described anywhere in the service, but the company does not say so; a plain statement that the service handles no protected health information, with its scope explained, is what would let a buyer rely on it.

Onboarding for clinical roles can involve immunization records, occupational health clearance or fitness for duty documents, which are health information; nothing says whether any of them pass through the platform.

The company names a recognized certification in its own material and publishes no artifact. The Lyn page states that Incredible Health is SOC 2 Type II certified and compliant with leading security, data and AI standards, and the site footer links a trust center hosted on Vanta. The site publishes no report, date or scope, though the trust center may hold the report.

Scope is the open point. The platform holds license numbers, certifications, employment histories and recorded interviews for a national population of nurses, and recruiter accounts at competing hospitals reach those profiles, so provisioning, scoping and revocation of recruiter access are the controls that matter.

No device claim is made and none applies: this is a hiring marketplace and interview agent, and nothing it sells touches patient care. What governs it is employment law, including rules on automated employment decision tools. The Lyn page says the agent is compliant with AI hiring laws across state and local jurisdictions, and the terms tell candidates that a human recruiter makes all final decisions; no bias audit summary, candidate notice text or allocation of compliance duties between the company and the employer is published.

Employment and AI hiring law is named here as what governs the product; it is not a device pathway.

Compliance is asserted and no evaluation is published. The Lyn page says the agent is compliant with AI hiring laws across state and local jurisdictions and is built to clear healthcare security, AI governance, legal and procurement reviews. No bias audit, adverse impact analysis, subgroup result or description of how the matching and interview models are tested is published on the company's pages, in the privacy policy or in the terms.

Bias is a pointed concern for this product, because the models decide which nurses an employer sees and how a candidate's phone interview is summarized. A matching model that learns from which candidates employers chose can reproduce the patterns in those choices, and proxies such as employment gaps, tenure or accent in a voice interview correlate with characteristics the law protects. New York City requires a published summary of an independent bias audit before an automated employment decision tool is used for roles in the city, and the marketplace serves New York employers; no audit summary is published by the company.

Review mechanisms exist and nothing stands behind the output. Lyn's interviews are recorded and transcribed, and recruiters receive a structured summary with the recording, so a person can listen to what the candidate said rather than rely on the summary; the terms state that a human recruiter reviews the results and makes all final decisions about a candidacy.

The terms then place accuracy on the candidate. Where a hospital's applicant tracking system feeds the platform, the company does not guarantee the imported data, the candidate assumes sole responsibility for verifying it during the interview and on the platform, and the company disclaims liability for employment decisions resulting from outdated, inaccurate or incorrectly mapped data. No error rate for matching or for Lyn's summaries is published, and no route is described for a nurse to see or contest how she was ranked or summarized; the privacy policy's access, correction and deletion rights reach personal data, not the model's output. Disputes go to individual arbitration unless the user opts out.

No clinical record integration exists: this is a hiring marketplace and nothing connects to an electronic health record. For this buyer, the systems of record that matter are an employer's applicant tracking and human resources platforms, and those are named. The Lyn page says the company is a Workday Innovation Partner approved for integration and that it integrates with major applicant tracking systems including Oracle, iCIMS and SAP, with a two week go-live; the marketplace takes an employer's requisition list or job feed. The terms describe the direction for candidates: where a hospital integrates its applicant tracking system, the company may automatically import a candidate's resume, history, contact details and application data from it.

What is not published is whether Lyn's summaries, match results and interview outcomes are written back into the employer's system, and which customers run each connection in production.

A single hosted service, with location implied rather than committed. The marketplace runs as a web and mobile service for United States employers and healthcare workers, and the privacy policy describes service providers who process data on the company's behalf without naming a host, a region or a residency commitment. Interview recordings and transcripts, profiles built partly from public licensure records, and data imported from employers' applicant tracking systems all rest somewhere unstated. The company does run a trust center on Vanta, which may name the host.

Tenancy matters commercially here, because the customers are hospitals competing for the same nurses in the same regional markets.

How the product is sold is published and no price is. The marketplace FAQ says the marketplace is an annual subscription customized to the employer's needs and size, and that list pricing is not published because enterprise contracts vary; a demo produces a quote. The service is free to nurses and healthcare workers, and the Lyn page describes the voice agent as an employer product with a two week go-live. That gives the unit of charge (an annual subscription) and the fact that size drives it, but not the figure.

Coverage is named precisely, with outcome figures behind part of it. The marketplace is for permanent roles, not travel or contract staffing, at hospitals, health systems, ambulatory surgery centers and home health organizations. It lists its specialties (ICU and critical care, medical-surgical, emergency, perioperative, NICU and PICU, home health, oncology, behavioral health, nursing leadership, labor and delivery, allied health) and names the states where the marketplace is not offered, with allied and non-clinical roles supported only in select geographies; Lyn is offered nationwide.

The published results cover acute care systems and ambulatory surgery centers (Johns Hopkins, Montefiore, Providence, Alameda) and nursing roles. Nothing is published for home health, behavioral health or allied roles, and nothing separates results by specialty beyond Alameda's share of hard-to-fill hires.