Draup

Company overview

Draup was founded in 2017 by Vijay Swaminathan and Vamsee Tirukkala, originally headquartered in Bangalore, India and now listing its primary office at The Woodlands, Texas, following a USD 20 million Series A from HKW announced in March 2022. The platform aggregates 1B+ professional profiles, 27K+ skills, 3,500+ roles and 6,100+ locations drawn from public datasets, professional networks, data brokers and government labour sources, and applies 150+ machine-learning models plus two agentic assistants — Curie for workforce strategy and Etter for AI-driven work redesign — across workforce planning, predictive skills architecture, university hiring, peer benchmarking and outbound candidate sourcing. Draup reports serving 300+ global enterprises including five of the Fortune 10, sells on a custom enterprise quote basis, and distributes data via REST API, SDKs, cloud data feeds, Model Context Protocol and 30+ HRIS/ATS connectors including Workday, SAP SuccessFactors and Greenhouse.

Regulatory exposure

Draup's talent-acquisition module produces "Ranked profiles" and exportable candidate shortlists that feed directly into customer ATSs, which places it plausibly within Annex III of the EU AI Act as a high-risk system used for recruitment and candidate filtering, and potentially within NYC Local Law 144 where those rankings substantially assist screening. The gap between Draup's claims and its published evidence is the central concern: the talent-acquisition page asserts that "Outputs comply with NIST AI RMF, ISO/IEC 42001, and EU AI Act standards, ensuring transparency in each hiring decision," yet Draup publishes no technical documentation, model card, conformity assessment, certificate or bias audit to substantiate any of it, and there is no ISO 42001 certification claim anywhere on the site. Its repeated line "As an EAIGG member, we audit for bias" refers to membership of the Ethical AI Governance Group, a 501(c)(3) practitioner community rather than an audit or certification body, and Draup does not appear on EAIGG's own site. There is no /security page (HTTP 404), no trust centre, no status page (status.draup.com does not resolve), and no NYC LL 144 or EU AI Act deployer guidance. Separately, the product exposes "Diversity filters (Ethnicity, Gender, Veterans)," a capability that deployers should scrutinise carefully under Title VII and EU non-discrimination law even when used for outbound sourcing rather than selection.

Path to a higher score

The fastest credibility win is to reconcile claim with evidence: either publish an ISO/IEC 42001 certificate and a NIST AI RMF mapping, or soften the "outputs comply with ... EU AI Act standards" assertion on the talent-acquisition page. Beyond that, Draup should commission and publicly post an independent bias audit of its candidate-ranking models in NYC LL 144 format — naming auditor, date, selection and scoring rates by sex and race/ethnicity, and intersectional categories — and repeat it annually; publish an Annex IV-style technical documentation pack or model card covering the ranking models, their training inputs, known limitations and accuracy metrics; add an Article 27 FRIA template plus explicit deployer-obligation guidance for EU customers; stand up a trust centre exposing the SOC 2 report and ISO 27001 certificate under NDA, a sub-processor register, a training-data exclusion list and a standard DPA; and add a public status page, a security.txt/vulnerability disclosure contact and a model-update changelog so post-market monitoring is externally visible.