Radancy

Company overview

Radancy is a New York-headquartered talent acquisition cloud that operated as TMP Worldwide Advertising & Communications before rebranding to Radancy in 2021, and has been backed by private equity (Gemspring Capital) since 2018. Its own site reports 675+ platform customers, roughly 965 employees and 18 offices across 10 countries, with named enterprise references including Capital One, Dell, Sony, UPS and Santander. The platform spans recruitment marketing, career sites, candidate CRM and analytics, and has been extended through acquisitions — Brazen and Asendify in 2023 and myInterview in September 2025, whose domain now redirects into Radancy's screening-and-scheduling product. Radancy markets five named agentic AI agents (Career Discovery, Sourcing, Screening, Scheduling and Insights) that match candidates to open roles, evaluate qualifications via conversational AI, and automatically score and rank candidates against job requirements using a proprietary skills knowledge graph.

Regulatory exposure

Radancy's exposure is high and broad. Automated scoring, ranking and shortlisting of candidates places the platform squarely in Annex III of the EU AI Act as a high-risk employment system, triggering Article 11 technical documentation and Article 27 deployer FRIA obligations — and Radancy is a genuine EU-market actor with offices in Germany, France, Austria and the Netherlands and localized Responsible AI pages in French and Dutch. In New York City, a tool that scores and ranks applicants to substantially assist hiring decisions is an AEDT under Local Law 144, requiring an annual independent bias audit with publicly posted summary results. The myInterview one-way video capability adds Illinois AI Video Interview Act duties, and Colorado SB 24-205 would treat Radancy as a developer of a high-risk AI system with attendant documentation obligations to deployers. Against that exposure, the most striking gap is evidentiary: despite explicitly selling AI candidate scoring and ranking, Radancy publishes no bias audit, no auditor name, no fairness testing results, and no jurisdiction-specific compliance guidance anywhere on its public site. Its most concrete commitments are contractual rather than technical — its standard Terms bind it to human decision-making and non-discriminatory AI design, which is meaningful but is not a substitute for a published audit or technical file.

Path to a higher score

The single highest-value step is commissioning and publicly posting an NYC Local Law 144 bias audit for the screening/scoring agent and the myInterview video product, with a named independent auditor, audit date, selection and scoring rates by sex and race/ethnicity including intersectional categories, and a downloadable summary — this alone would move the weakest score dramatically. Second, publish Article 11-grade technical documentation: a model or system card for the skills knowledge graph and scoring model, an explainability statement making good on the 'explainable scoring' product claim, and a description of intended purpose, known limitations and accuracy metrics. Third, convert the DPA's existing DPIA-assistance clause into explicit EU AI Act deployer guidance, including an Article 27 FRIA template and an instructions-for-use pack. Fourth, pursue ISO/IEC 42001 certification and upgrade the DPA's 'ISO/IEC 27001 alignment' language to an actual certification, publishing both alongside the existing SOC 2 Type 2 + HITRUST report rather than gating them behind a request form. Finally, stand up public post-market surveillance surfaces — a status page (status.radancy.com does not currently resolve), an ungated model-update changelog, and a named AI incident channel distinct from the existing security-concern form.