Talogy

Company overview

Talogy is the assessment brand created in 2022 when PSI Talent Management consolidated seventeen acquired firms — including Cubiks, Caliper, Select International, a&dc and IPAT — under a single identity; its own About page traces the lineage to 1946 and psychologist Floyd E. Ruch. The company is headquartered at 611 N. Brand Blvd in Glendale, California, operates in 25+ countries with 750+ staff, and sits under parent holding company Lifelong Learner Holdings LLC, which is named as data controller in Talogy's Research Data Privacy Statement. Its TalogyTech platform delivers cognitive, personality, values and leadership assessments with configurable reporting and integrations into Taleo, Oracle, Workday, Kenexa, iCIMS and SAP SuccessFactors. Pricing is not published and product use generally requires accreditation, indicating an enterprise, consultative sales motion.

Regulatory exposure

Talogy is squarely exposed as an AI-system provider under EU AI Act Annex III(4): its tools evaluate candidates for recruitment and selection, and it markets machine-learning-derived scoring in the Mindgage battery. It is also plainly in scope for NYC Local Law 144 wherever its scored assessments substantially assist hiring decisions in New York City, and for Illinois and Colorado obligations. The gap between that exposure and its public record is wide. Talogy publishes a seven-point AI ethics framework and cites the OECD principles, the NIST AI Risk Management Framework and SIOP 2023 validation guidelines, but it publishes no bias audit by any independent auditor, no summary of results, no model or system card, no instructions for use, no ISO 42001 or SOC 2 attestation, and no FRIA or deployer guidance of any kind. Its fairness claims — up to a 60% reduction in subgroup differences for Mindgage — are self-reported and described in its own blog as 'preliminary evidence' from a pilot study, with no statistics, effect sizes or methodology disclosed. Its own ethics article concedes there are 'not yet established best practices' for demonstrating ethical compliance, which is a fair description of its own disclosure posture. Its data-protection documentation is genuinely above average — a detailed privacy policy, a named sub-processor list with locations, an EU-U.S./UK/Swiss Data Privacy Framework certification and a separate research data statement disclosing R&D reuse of assessment data including optional ethnicity — but that is privacy compliance, not AI-governance compliance.

Path to a higher score

The single highest-value move is to commission and publish a downloadable independent bias audit of the scored assessments most used in US hiring, naming the auditor, the date, the sample and intersectional impact ratios, and to repeat it annually. Second, convert the internal validation work that already exists behind accreditation into public technical documentation: a per-instrument technical manual or model card covering intended purpose, construct, training and norming data, known limitations, and an explainability statement that substantiates the 'humanly interpretable' claim made for Mindgage's hybrid scoring. Third, publish the underlying statistics behind the 60% subgroup-difference claim rather than leaving it as an unsourced marketing figure. Fourth, add an AI-specific section to the legal hub carrying deployer guidance for EU AI Act Article 26/27, NYC LL 144 notice templates, and Illinois and Colorado obligations, alongside a public DPA. Finally, stand up basic post-market surface — a security contact and vulnerability disclosure route, a model and scoring changelog, and a stated cadence for re-checking adverse impact in live client data. ISO 42001 certification would consolidate several of these at once.