HireVue

Hirevue publishes a Certified Integration Partner Program with twelve named applicant tracking systems, several carrying partner tier and live customer counts. Workday is a Select Partner with a certified prebuilt integration and 140+ live customers. SAP is a Silver Partner covering SuccessFactors Recruiting, with 50 live customers and a listing on the SAP Store.

Oracle is a Gold Partner covering ORC and Taleo, with over 100 live customers on the Cloud Marketplace. SmartRecruiters is a certified partner with self service activation through its Marketplace. PageUp, iCIMS, Avature, Infinite for BrassRing, Greenhouse, Oleeo and Cornerstone with Talentlink complete the list.

Eightfold adds a talent intelligence integration, and Deloitte, AMS and TalentA are named implementation partners.

The integrations act rather than just list. The AI Interviewer FAQ states results flow back to the candidate record so recruiters and managers stay in their normal tools, the platform integrates with 45+ ATSs with detailed data where the customer needs it, and scheduling automation syncs recruiter calendars. Those are authenticated writes into systems Hirevue does not own, and the live customer counts are the vendor's own evidence that they run in production.

One limit, which does not affect the grade: the integrations center on the ATS, so the breadth is depth in one ecosystem plus calendar and talent intelligence.

Sourcehirevue.com/platform/integrations and /platform/ai-interviewerread 2026-09-08

Stages do chain and transitions are automated. Workflow Automation is a named module in the primary product navigation, and the AI Interviewer page states the product "integrates conversations and assessments to stack signals, while automated workflows smooth transitions in the hiring process," with results flowing back into the ATS candidate record.

The agent's own sequence is documented in four steps: Configure, Consent, Converse, Evaluate. The customer sets questions and picks from a skills bank, the candidate consents, the AI conducts a two way voice conversation that asks, probes and listens, and then it scores each answer against the rubric and explains why.

It is not Full because the Workflow Automation module resolves to interview scheduling software, not a general orchestration builder. The chaining runs between Hirevue's own products (Talent Engagement Tools, AI Hiring Agents, Assessments, AI Interviewer, Interview Insights) in a sequence Hirevue defines rather than one the customer authors. Nothing documents branching, conditional routing, handoff between the named agents, or a multistep flow the customer configures. A fixed vendor pipeline is Partial.

Sourcehirevue.com/platform/ai-interviewer and /platformread 2026-09-08

Hirevue grounds its scoring in a large, maintained library on the vendor side: decades of validated industrial organizational psychology, built on 70M+ validated interactions, with an "IO-validated skills bank" a customer selects from when configuring an interview. That library persists between runs and scales, and it is real grounding rather than a generic model, which is why this is not None.

It is not Full because the bar asks for a maintained retrieval structure over the customer's own knowledge, and no customer specific retrieval or document grounding is documented. What the customer contributes is configuration: screening questions, competencies, pass criteria and a rubric set per role, level, location and team, assembled per interview.

Nothing documents ingestion of the customer's job descriptions, competency frameworks, internal documentation or past scorecards as a grounding layer the agent queries. The promise to ask "the same questions of all candidates" points away from per customer retrieval rather than toward it.

Sourcehirevue.com/platform/ai-interviewer and /our-scienceread 2026-09-08

Hirevue documents three real controls. Candidates review and agree to AI assisted recording, evaluation and scoring before the interview begins, which is a precondition on the agent running at all, though it faces the candidate rather than giving the customer an oversight surface. The product page names "configurable governance controls" alongside bias mitigated scoring and audit ready workflows. And the customer sets the boundary the agent scores against: questions, competencies and pass criteria are configurable per role, level, location and team, calibrated with hiring managers.

It is not Full because none of that is a review and approve surface over what the agent does. The agent interviews, scores against the rubric and produces a "recruiter-ready shortlist by morning," with no approval queue, hold, confidence threshold or reversal route documented before an output reaches a recruiter.

Hirevue positions agents to take the high volume work while hiring decisions stay with people, and standardized rubrics enforce consistency. Reserving the decision is product posture, and a rubric is a scoring instrument, not an oversight mechanism. The agent recommends rather than acts, so the missing gate is partly structural, but structure is not a mechanism.

Sourcehirevue.com/platform/ai-interviewerread 2026-09-08

Hirevue documents both halves: independent attestation and access the customer administers by role.

On attestation, Hirevue is certified to ISO/IEC 27001:2022 for information security management and ISO/IEC 27701:2019 for privacy information management. The stated scope covers the cloud hosting environment, personnel, policies and procedures, and the systems and networks delivering the service, across both the VIP and T2O solutions.

An annual SOC 2 Type 2 covers all five Trust Services Criteria (Security, Availability, Processing Integrity, Confidentiality and Privacy), again across both solutions. Hirevue also has a SOC 3 report and EU-U.S. Data Privacy Framework participation, and publishes a Trust Center at trust.hirevue.com for controls documentation. Hirevue claims to be the only video interviewing vendor third party certified for both application and data storage.

On access, a Detailed Permissions surface lets a customer grant access to the data a role needs and withhold data irrelevant to that role, with the roles named as recruiter, coordinator, hiring manager and interviewing panelist. Single sign on is available for enterprise deployment.

The FedRAMP authorization is counted under deployment as a named customer environment, and security is Full without it.

Sourcehirevue.com/platform/enterprise-security-compliance and trust.hirevue.comread 2026-09-08

Hirevue documents a per decision audit trail by name. The Evaluate step of the AI Interviewer's four step flow reads "Full audit trail included. AI compares each answer to your rubric, scores it, and explains why." That is a record of each decision with the rubric criterion, the score and the reasoning attached.

Three more statements on the same page back it. The AI Interviewer is built for trusted, transparent hiring with "bias-mitigated scoring, audit-ready workflows, and configurable governance controls." It delivers "explainable, audit-ready decisions." And the FAQ states that a structured interview documents "the why" and lets a customer "monitor outcomes over time." The consent step adds a dated record that the candidate reviewed and agreed to AI assisted recording, evaluation and scoring before the interview began.

Explaining a score is not, on its own, a way to reconstruct what an agent did. Hirevue clears that line because it documents the trail itself as a shipped inclusion, not as an inference from explainability. The bias audit program is covered under testing and the detailed permissions surface is counted under security, so neither is counted here.

Sourcehirevue.com/platform/ai-interviewerread 2026-09-08

No agent memory layer, cross session state store, retained preference set, or memory scope and lifetime is documented on the AI Interviewer page, the security page, the integrations page or the site navigation.

Interview recordings, transcripts, scores and audit trails persist against the candidate record and flow back into the customer's ATS. That is the application's own data: deleting it deletes the hiring record itself, so it does not count as agent memory. The audit trail is counted under observability.

The product design points away from memory. A structured interview asks "the same questions of all candidates," so consistency between runs is what is being sold, and an agent carrying state from one candidate's interview into the next would undermine it. The AI asks, probes and listens, following the answer given, which is context within a single run rather than persistence across runs.

Partial would need a stated memory scope, a lifetime, or a retention or deletion control over agent memory as distinct from candidate data, and none is published. The customer can monitor outcomes over time and tune the rubric, but that is configuration the customer changes, not state the agent retains, and it is counted under testing.

Sourcehirevue.com/platform/ai-interviewer and /platform/enterprise-security-complianceread 2026-09-08

Hirevue states it is "co-located in the U.S. and Europe" to serve clients in 180+ countries, which is a named region list.

Hirevue also runs a FedRAMP authorized environment for federal hiring, audited by a third party assessment organization and authorized by the FedRAMP PMO, so federal hiring can use tools proven in the private market. That is a separately assessed environment that public sector customers are placed into, distinct from the commercial platform. Either the two named regions or the named customer environment would support Full on its own.

The gap is that no selection surface is published. Nothing states how a customer elects EU hosting, whether it follows from entity location, is a contract term, or is chosen at provisioning. No VPC, on premises or customer tenant option is documented, though Full does not require one. The FedRAMP authorization is counted here as a deployment environment and not under security, which stands without it.

Sourcehirevue.com/platform/enterprise-security-complianceread 2026-09-08

Hirevue ships a set of separately paged products: a Skill Validation family of Virtual Job Tryout, Game-Based Assessments, Technical Assessments and Language Proficiency Tests; an Assessment Builder for composing new ones; an AI Hiring Agent; an AI Interviewer; Video Interviewing; Interview Insights; Talent Engagement Tools; Match and Apply; and Workflow Automation.

Beneath them sits a library of validated assessments, structured interview templates and an "IO-validated skills bank" a customer chooses from when configuring a role. That library is a real reusable template bank, which is why this is not None.

It is not Full because it is a catalog of assessment content, not of agents. The skills bank supplies competencies to score against and the assessment library supplies instruments to run; neither is a set of complete working agents a buyer selects among. The products above are modules of one platform, which the bar treats differently from a pack. No agent gallery, clone and configure surface or marketplace of Hirevue authored agents appears anywhere. The Certified Integration Partner Program lists third party connectors and is counted under integrations.

Sourcehirevue.com/platform/assessment-builder, /platform and /platform/ai-interviewerread 2026-09-08

Channels are broad. The AI Interviewer runs 24/7 two way voice interviews in which the AI asks, probes and listens. On demand and live video interviewing are separate named products, conversational AI engages candidates through Talent Engagement Tools and career site conversation, and scheduling automation removes the booking exchange. Availability is the product's core claim: a candidate interviews on their own time and recruiters get shortlists by morning, with an 86% completion rate cited.

It is not Full because the trigger side is thin. What is documented is candidate initiated conversation, recruiter or ATS driven invitation, and automated transitions between Hirevue's own stages. No event framework, webhook, subscription surface or customer configurable trigger on business events is published, so the agent does not start work off a change in the customer's systems. Two of the four markers the bar names are present: the channel spread is wide while the event surface is narrow.

SMS and WhatsApp invites and text to apply are not counted, because none of the pages cited shows them; hirevue.com/platform/candidate-engagement-tools could confirm them.

Sourcehirevue.com/platform/ai-interviewer and /platformread 2026-09-08

The customer has no choice over the models running the AI Interviewer. No model picker, admin setting, provider list, routing description or bring your own model option appears on the AI Interviewer page, the security page, the integrations page or the site navigation, and Hirevue names no model provider for its scoring or conversational AI.

What Hirevue does disclose is how its own models are built. It publishes an AI Explainability Statement, which it describes as the first in the HR industry and which was reviewed by the UK Information Commissioner's Office.

The statement covers how its AI based assessments are built, tested and validated, including multipenalty optimized models and ongoing bias mitigation, and Hirevue open sources techniques such as algorithmic debiasing. That is unusual depth on the vendor's own models, but disclosure of what the vendor built is not customer choice over what runs. The AI Ethical Principles and the compliance alignment list address governance, not which models run.

The full 2024 Explainability Statement PDF and legal.hirevue.com/product-documentation could name a provider, and a disclosed routing choice there would make this Partial.

Sourcehirevue.com/platform/ai-interviewer and /ai-in-hiringread 2026-09-08

Hirevue runs a Certified Integration Partner Program with its own Become a Partner page, under which third parties build and certify integrations against Hirevue. It also keeps vendor written listings on three marketplaces: the SAP Store, the Oracle Cloud Marketplace and the SmartRecruiters Marketplace.

The SmartRecruiters integration offers self service activation, so a customer connects the two systems without a services engagement. The Talent Engagement Agent is published into the Workday Agent Partner Network and connected to the Workday Agent System of Record.

A partner program with certified third party builds and self service activation suggests a platform interface behind it, which is why this is not None.

It is not Full because the bar asks for a documented API or SDK for the vendor's own platform. No developer portal, API reference, endpoint list, authentication guide, SDK or MCP server appears on the platform pages, the integrations page, the security page, the site footer or the legal center at legal.hirevue.com. developers.hirevue.com and docs.hirevue.com could hold a published reference, which would make this Full.

Sourcehirevue.com/platform/integrations and /platform/integrations/partner-programread 2026-09-08

The customer has a tuning loop. The FAQ states that a customer can "monitor outcomes over time and tune the rubric," and questions, competencies and pass criteria are configurable per role, level, location and team, calibrated with hiring managers. That is adjustment against observed output on the customer's own configuration, which is Partial.

Hirevue also runs a large audit program on its own algorithms. It has voluntarily commissioned independent audits, engaged DCI Consulting Group as an external bias auditor, and publishes an AI Explainability Statement, which it describes as the first in the HR industry and which the UK Information Commissioner's Office reviewed. Each of those tests Hirevue's own models in general and is published as a disclosure, so it does not count here. The size and quality of the program do not change that; what counts is whose deployment was tested.

Alignment with UGESP, EEOC, ADEA, ADA and emerging algorithmic fairness guidelines does not say whose deployment was tested either. Hirevue's own press material notes that NYC Local Law 144 places audit responsibility on the employer. A customer side audit of that kind would count, but Hirevue does not document supplying one.

It is not Full because no test environment, holdout, simulation or scored test case run before deployment is published for the customer's configured interview.

Sourcehirevue.com/platform/ai-interviewer, /ai-in-hiring and /press-release DCI Consultingread 2026-09-08

No hosted or local browser, desktop session, or remote or local computer control by the agent is documented on the AI Interviewer page, the platform navigation, the integrations page or the security page.

The near miss on a video interviewing platform is the candidate's own device. The candidate's webcam and browser are where the interview happens, and on demand video interviewing records a candidate through their own device. That is how the product reaches a human participant, not the agent driving an interface. The AI asks, listens and scores a conversation; it does not navigate, click or operate anything on anyone's behalf.

Technical Assessments run candidate code, and sandboxes, terminals and code execution do not count as computer use. The usual buyer question here, how often it breaks when UI elements change, does not apply to a voice conversation scored against a rubric. No sourcing or prospecting browser extension appears anywhere in Hirevue's products either.

Sourcehirevue.com/platform/ai-interviewer, /platform and /platform/integrationsread 2026-09-08