Cognisess
HireAIScore rates Cognisess 30 out of 100 (grade F, Substantial gaps) under rubric v1.0, last reviewed September 25, 2026. Cognisess ranks 82 of 82 vendors rated in Assessment AI, against a category average of 46. That score is drawn from 26 evidence items across 7 rubric criteria for Cognisess — 21 with a cited source, 5 recording that nothing was located.
Score breakdown
Cognisess, criterion by criterion
Each rubric criterion is scored out of 100 on the cited evidence, then weighted into the total.
Weighted contribution shown to the right of each bar.
Article 11 Technical Documentation
Weight 20% · category avg 50
25
+5.0
Bias Audit Transparency
Weight 18% · category avg 41
20
+3.6
FRIA Support
Weight 15% · category avg 31
20
+3.0
Data Governance Disclosure
Weight 15% · category avg 53
42
+6.3
Human Oversight Design
Weight 12% · category avg 55
42
+5.0
Post-Market Monitoring
Weight 12% · category avg 38
32
+3.8
Customer Documentation
Weight 8% · category avg 58
40
+3.2
Total weighted score
30.0(rounded to 30 for display)
Category avg is the mean raw score on that criterion across the 82 Assessment AI vendors in scope of this rubric, this one included.
Strongest criterion
Human Oversight Design
42raw / 100
Contributes 5.0 to the total. 42 against a 55 category average.
Weakest criterion
FRIA Support
20raw / 100
Contributes 3.0 to the total. 20 against a 31 category average.
Cited evidence
The evidence behind the score
Every score is backed by at least one cited piece of evidence. Open a criterion to see what was found, when, and where.
Download diligence recordEvidence ledger
- Items
- 26
- Documentation
- 14
- Public statement
- 7
- Absence
- 5
- With a source URL
- 21 of 26
- Source hostnames
- 6
- Fewest items
- 3
- Article 11 Technical Documentation
These figures measure how thoroughly Cognisess was reviewed, not how Cognisess performed — an absence row, recording that nothing was located, is counted like any other item.
Article 11 Technical Documentation
3 items25
- AbsenceCaptured Sep 25, 2026
No model card, technical file, instructions for use, explainability statement or ISO/IEC 42001 certificate is published: the relaunched cognisess.com sitemap has ten URLs of marketing and legal pages, the Privasee privacy portal has none, and the 'Ethical AI statement' cited in the data protection policy is not linked anywhere.
- DocumentationCaptured Sep 25, 2026
The live data protection policy is the only current description of the AI; it says processing covers 'the calculation of job profiler scores, the calculation of inferred variables and attribute values, and the work of the AI models that comprise our video analytics tool', and gives no model logic, inputs or limitations.
- Public statementCaptured Sep 25, 2026
The Ethical AI page (published 2022, last archived 20 January 2026, now 404) offered only principles, such as models 'tested for adverse impact' and protected characteristics never used, and pointed to a 'full Ethical AI policy' that was never linked.
https://web.archive.org/web/20260120095515/https://cognisess.com/ethical-ai-2/
Bias Audit Transparency
3 items20
- AbsenceCaptured Sep 25, 2026
No independent bias audit, adverse-impact statistics or NYC LL 144 summary is published on cognisess.com, in its privacy portal, on its archived pages or on any third-party audit dashboard checked (trust.warden-ai.com/cognisess returns 404).
- Public statementCaptured Sep 25, 2026
A July 2019 blog post, still live in February 2026, claimed the video tool 'has analysed over 1.4 billion facial expressions with a 97% accuracy rate' and 'isn't programmed to see race, age or gender', supported only by an unquantified single-customer correlation with assessor ratings.
- Public statementCaptured Sep 25, 2026
The About page says its data is 'All validated with live benchmarking across education, employment, and leadership contexts' and sells 'fairness audits' as a consulting service, but publishes no validation study, impact ratios or named auditor.
FRIA Support
3 items20
- AbsenceCaptured Sep 25, 2026
Cognisess publishes no EU AI Act material (no FRIA template, no Art. 26/27 deployer guidance, no instructions for use, no warning about the Art. 5(1)(f) emotion-recognition ban) on its website, in its legal documents, in its privacy portal or in its archived knowledge base.
- DocumentationCaptured Sep 25, 2026
The acceptable use policy only asks that 'Results are interpreted responsibly' and makes users responsible for legal compliance, with no regulatory guidance.
- Public statementCaptured Sep 25, 2026
Legal commentary on the Art. 5(1)(f) workplace ban (Wolters Kluwer Global Workplace Law & Policy, Stefania Marassi, 14 April 2025) names Cognisess among vendors claiming to read employees' emotional and mental states from biometric data, and Cognisess has published no response or deployer guidance.
Data Governance Disclosure
5 items42
- DocumentationCaptured Sep 25, 2026
The data protection policy says Cognisess 'does not scrape data from the Internet'. It collects demographic data optionally for bias monitoring, hosts data on Azure in the UK, Europe, Australia and China, deletes accounts inactive for two years while keeping anonymised scores for norming, and hides personality answers and undisclosed disabilities from employers.
- DocumentationCaptured Sep 25, 2026
The DPA lists 'Self-recorded audio and video files' and 'biometric / bio-mechanical performance data' (heart rate, EEG, ECG) among processed data, but its transfer terms are out of date: they cite Privacy Shield, the 2004/2010 SCCs and Directive 95/46/EC, and describe GDPR as becoming enforceable on 25 May 2018.
- DocumentationCaptured Sep 25, 2026
The data behind Cognisess's Privasee privacy notice (linked from its footer as the Privacy Policy) names DPO Phil Greening and ICO registration ZA290016 and lists Azure (EEA, Australia) and Microsoft authentication (US) as sub-processors. It declares no biometric data and no automated decision-making, which contradicts the DPA and the video-analytics feature.
https://products.privasee.io/api/privacy-portal/655f5b9b805bda0013437200
- Public statementCaptured Sep 25, 2026
The Video Analyser product page, live until at least 18 February 2026 and now removed, said 'Our AI driven video analytics tool measures facial expressions and language' and that Deep Learn can 'measure different facets of behaviour and emotion based on facial expressions and language used', without saying how that facial data is governed.
https://web.archive.org/web/20260218000400/https://cognisess.com/video-analyser/
- DocumentationCaptured Sep 25, 2026
The pre-relaunch homepage showed a UKAS ISO 27001 mark alongside 'Certificate Number 20831' (the current footer shows only the number), but the certifying body, scope and validity are not published.
https://web.archive.org/web/20260306201220/https://cognisess.com/
Human Oversight Design
5 items42
- Public statementCaptured Sep 25, 2026
The policy promises: 'We do not fully automate any hiring campaign, so there will always be a human making the decision about who moves onto the next stage of hiring.'
- DocumentationCaptured Sep 25, 2026
The live app's interface text contradicts that promise: its 'Automatic streaming' option says 'Candidates will be automatically streamed using the thresholds without any human interaction', and it sits beside a human 'Video appraisal' option.
https://cognisesscdnprod.azureedge.net/cognisess-4-29-0-392/public/locales/WebResources.json
- DocumentationCaptured Sep 25, 2026
The now-removed knowledge base documented manual, semi-automatic and automatic auto-stream modes and warned that 'all candidates below the fail threshold will be rejected'.
- DocumentationCaptured Sep 25, 2026
The knowledge base also documented an admin activity log that provides 'an in-platform audit trail of admin activity' for compliance or support investigations.
- Public statementCaptured Sep 25, 2026
The video-analysis blog sold the tool on the basis that 'there isn't any human involvement in the analysis process'.
Post-Market Monitoring
3 items32
- DocumentationCaptured Sep 25, 2026
The terms, last updated 27 November 2017, give security@cognisess.com as the security contact, promise to notify subscribers of security flaws, and guarantee 99% uptime measured on the login page by an external monitor.
- DocumentationCaptured Sep 25, 2026
status.cognisess.com is a StatusCake public report titled 'Cognisess Status' that lists no monitored services, uptime figures, announcements or incident history.
- AbsenceCaptured Sep 25, 2026
Cognisess publishes no model-update changelog, incident disclosures, bias-drift monitoring results or security.txt (www.cognisess.com/.well-known/security.txt returns 404), and the archived Ethical AI page's promise to routinely re-check its models disappeared in the 2026 relaunch.
Customer Documentation
4 items40
- DocumentationCaptured Sep 25, 2026
Current product pages are short marketing lists such as '50+ Enterprise Assessments', 'Automated Candidate Shortlisting', 'Predictive Performance Scoring' and 'DE&I Tracking & Dashboards', with no methodology, no video-analysis disclosure and no compliance guidance.
- DocumentationCaptured Sep 25, 2026
The live app's interface text still offers a 'Video analytics' setting ('Get extensive analysis of user videos. Contact us at support@cognisess.com to enable this feature.'), a 'Video analytics report' on 'the video analytics that DeepLearn™ can provide' and a 'Facial expression' section, none of which the public website discloses.
https://cognisesscdnprod.azureedge.net/cognisess-4-29-0-392/public/locales/WebResources.json
- DocumentationCaptured Sep 25, 2026
Until early 2026 a GDPR guidance page for clients covered lawful bases, data minimisation through staged assessment, Azure data residency and fully anonymised 'blind' recruitment, but it was removed in the relaunch along with the knowledge base.
https://web.archive.org/web/20260120083659/https://cognisess.com/gdpr/
- AbsenceCaptured Sep 25, 2026
The live site has no FAQ, resource hub, candidate-notice template or NYC LL 144 / EU AI Act guidance, and the terms still point customers to a data-processing addendum at 'cognisess.wpengine.com/terms-of-use' that returns 404.
Editorial notes
Company overview
Cognisess Limited (UK company 07505129, incorporated January 2011) is a small talent-analytics vendor based at Tramshed Tech in Newport, Wales. It is part of the DeepLearn Human Science Group ('A DLHS Company'), which also runs the Yondur careers platform, and its controlling shareholder is director Christopher Butt. The platform scores candidates and employees with 50+ assessments (games, questionnaires and cognitive tasks; older material cites 60+), weighted job-fit 'profilers' and a built-in ATS that can auto-stream applicants. It also runs 'Ask Me'/'Meet Me' recorded video interviews with optional DeepLearn video analytics. Named clients include IHG, AB InBev, Coca-Cola, DPD, Adecco and Waitrose. A 2026 relaunch cut the website to a few marketing and legal pages and dropped its Video Analyser, Ethical AI, GDPR-guidance and knowledge-base pages.
Regulatory exposure
The most urgent risk is the EU AI Act ban on AI that infers emotions in the workplace (Art. 5(1)(f)), which has applied since 2 February 2025. Cognisess's Video Analyser page was still live on 18 February 2026. It said the tool 'measures facial expressions and language' and uses Deep Learn to 'measure different facets of behaviour and emotion based on facial expressions'. A companion blog offered to 'detect a candidate's true emotion'. The page disappeared in the 2026 relaunch, but the live data protection policy still refers to 'the AI models that comprise our video analytics tool'. The production app's interface text still offers video analytics on request, with a 'Facial expression' section. Cognisess's privacy portal says it operates in the EU, yet it has announced no withdrawal of the feature and no EU carve-out. Its scoring, shortlisting and threshold-based auto-rejection count as high-risk under Annex III point 4, with obligations due 2 December 2027. In the UK, UK GDPR covers the biometric and special-category data its DPA lists. No NYC LL 144 bias audit is published, so NYC employers would have to commission their own. For US use of its video analysis and automated rejections, Illinois' AI Video Interview Act and HB 3773, Colorado's SB 26-189 (from 1 January 2027) and California's FEHA automated-decision rules would also apply.
Path to a higher score
Cognisess should first say publicly whether its facial-expression and emotion video analytics has been retired or blocked for EU and workplace use. If it has, the feature should be removed from the product, the DPA and the privacy policy. Cognisess should then: publish a model card and instructions for use covering its assessments, profilers and auto-streaming, with validity evidence and adverse-impact statistics; commission and publish an independent bias audit with LL 144-style impact ratios; require human review of threshold-based rejections, so its promise of 'a human making the decision' holds; replace its 2017 terms and 2018-era DPA (which cite Privacy Shield and superseded SCCs); and restore its ethics and help content, together with EU AI Act deployer guidance, a status page that actually monitors services, and a model changelog.
Regulatory frame
What applies to assessment AI.
Assessment is the most-regulated surface in AI hiring. NYC LL 144, EEOC and OFCCP frameworks, Illinois HB 3773, and EU AI Act Annex III §4 all apply, with overlap on bias-audit obligations.
Compare
Cognisess against its nearest-scoring peers.
The closest scores in Assessment AI, side by side.
- Side-by-sideCognisess vs HireQuotient34F
- Side-by-sideCognisess vs Pulsifi33F
- Side-by-sideCognisess vs Arena Analytics32F
- Side-by-sideCognisess vs Intervue.io31F
Ranking
Others rated in Assessment AI
Ranked 82 of 82 by weighted total under rubric v1.0. The ordering is arithmetic on the rubric and carries no view on which tool suits a given hiring process.
- #79Pulsifi33F
- #80Arena Analytics32F
- #81Intervue.io31F
- #82CognisessThis profile30F
Conflicts of interest
No vendor pays for placement, scoring, or removal. Casework - the consulting firm that operates this directory - provides paid services to some vendors. Any active or recent (within 24 months) commercial relationship is disclosed on the affected vendor profile and the review is reassigned to an independent reviewer. See the full policy on About.
Casework has no commercial relationship with this vendor.