Textkernel
HireAIScore rates Textkernel 51 out of 100 (grade F, Substantial gaps) under rubric v1.0, last reviewed June 7, 2026. Textkernel ranks 24 of 86 vendors rated in Candidate sourcing AI, against a category average of 46. That score is drawn from 18 evidence items across 7 rubric criteria for Textkernel — 13 with a cited source, 5 recording that nothing was located.
Score breakdown
Textkernel, 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 48
55
+11.0
Bias Audit Transparency
Weight 18% · category avg 40
55
+9.9
FRIA Support
Weight 15% · category avg 31
32
+4.8
Data Governance Disclosure
Weight 15% · category avg 53
55
+8.3
Human Oversight Design
Weight 12% · category avg 56
60
+7.2
Post-Market Monitoring
Weight 12% · category avg 40
40
+4.8
Customer Documentation
Weight 8% · category avg 57
62
+5.0
Total weighted score
50.9(rounded to 51 for display)
Category avg is the mean raw score on that criterion across the 86 Candidate sourcing AI vendors in scope of this rubric, this one included.
Strongest criterion
Customer Documentation
62raw / 100
Contributes 5.0 to the total. 62 against a 57 category average.
Weakest criterion
FRIA Support
32raw / 100
Contributes 4.8 to the total. 32 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
- 18
- Documentation
- 10
- Audit report
- 1
- Public statement
- 2
- Absence
- 5
- With a source URL
- 13 of 18
- Source hostnames
- 3
- Fewest items
- 2
- FRIA Support
These figures measure how thoroughly Textkernel was reviewed, not how Textkernel performed — an absence row, recording that nothing was located, is counted like any other item.
Article 11 Technical Documentation
3 items55
- DocumentationCaptured Jun 7, 2026
Responsible AI page sets out a 'white box AI' / 'AI driven by humans' position and states solutions 'indicate exactly which criteria are used to construct the match', with a downloadable 'Reduce Bias in AI Recruitment' whitepaper.
- DocumentationCaptured Jun 7, 2026
Tx Platform developer documentation describes the matching engine's scoring (weighted score, Reverse Compatibility Score, SovScore) and per-category breakdowns, partially documenting how the AI reaches its output.
https://developer.textkernel.com/tx-platform/v10/search-match/overview/querying/
- AbsenceCaptured Jun 7, 2026
No system/model cards, formal Article 11 technical documentation pack, instructions-for-use, or ISO 42001 certification were found on textkernel.com, the developer portal, or the Responsible AI page.
Bias Audit Transparency
3 items55
- Audit reportCaptured Jun 7, 2026
Blog (dated Feb 2025) states Textkernel's Search & Match system was audited by Holistic AI for NYC Local Law 144, claiming impact ratios across gender and ethnicity consistently above the 0.8 benchmark.
https://www.textkernel.com/learn-support/blog/textkernel-passed-nyc-ai-law-audit/
- Public statementCaptured Jun 7, 2026
AI legislation blog references the company's NYC AEDT audit but links only to the summary post, not a downloadable bias-audit report.
- AbsenceCaptured Jun 7, 2026
No publicly downloadable Holistic AI audit report (with intersectional impact-ratio tables), AEDT candidate notice, or second-year/continuous bias audit was found on textkernel.com or the ACLU LL144 tracker.
FRIA Support
2 items32
- Public statementCaptured Jun 7, 2026
Blog explains the EU AI Act's high-risk classification and conformity-assessment routes for recruitment AI but does not cover Article 27 Fundamental Rights Impact Assessments or provide deployer templates/checklists.
- AbsenceCaptured Jun 7, 2026
No FRIA template, Article 27 deployer guidance, or fundamental-rights-impact-assessment material was found anywhere on textkernel.com or its policies/blog pages.
Data Governance Disclosure
2 items55
- DocumentationCaptured Jun 7, 2026
Developer FAQ states indexed parsed documents are scrubbed of all personal data/PII and that LLM Parser inputs and outputs are 'not stored, not used to train future models by Textkernel or AWS, and never shared' with outside parties.
- AbsenceCaptured Jun 7, 2026
No detailed training-data provenance/exclusion list, ISO 42001 certification, or publicly downloadable SOC 2 report was found; data-governance claims are high-level and certificates appear gated behind sales/NDA.
Human Oversight Design
3 items60
- DocumentationCaptured Jun 7, 2026
Source & Match product page states 'our Matching is not a black box, so recruiters can retain full control of the search' and promotes 'White box matching' where users 'remain in complete control of their search query from beginning to end'.
https://www.textkernel.com/products-solutions/source-and-match/
- DocumentationCaptured Jun 7, 2026
Developer docs show matches return scores broken down by category (Certifications, Education, Job Titles, Languages, Management Level, Skills, Industries, etc.) with suggested weights, giving reviewers an inspectable per-criterion rationale.
https://developer.textkernel.com/tx-platform/v10/search-match/overview/querying/
- DocumentationCaptured Jun 7, 2026
Responsible AI page frames the product as 'AI driven by humans' with users retaining decision-making authority and systems disregarding gender, ethnicity and age.
Post-Market Monitoring
2 items40
- DocumentationCaptured Jun 7, 2026
Responsible Disclosure Policy lists a security contact (security@bullhorn.com), a 2-business-day response commitment, a defined scope of products, and financial rewards for valid vulnerability reports.
https://www.textkernel.com/policies-statements/responsible-disclosure-policy/
- AbsenceCaptured Jun 7, 2026
No public continuous-monitoring dashboard, model-performance/drift monitoring page, model-update changelog, or status page was found for the AI matching/parsing products.
Customer Documentation
3 items62
- DocumentationCaptured Jun 7, 2026
Public product pages document the Source & Match sourcing/matching solution, its white-box approach, and bias-mitigation features.
https://www.textkernel.com/products-solutions/source-and-match/
- DocumentationCaptured Jun 7, 2026
Policies & Statements hub centralizes the privacy statement, GDPR/CCPA materials, responsible disclosure policy, and a Data Processing Agreement for deployers.
- DocumentationCaptured Jun 7, 2026
Customer-facing blog explains EU AI Act and NYC LL 144 obligations for recruitment AI and links to a Responsible AI webinar, though without downloadable compliance templates.
Editorial notes
Company overview
Textkernel B.V., founded in 2001 in Amsterdam as a spin-off of NLP/machine-learning research at Dutch and Belgian universities, builds AI for the HR-tech supply chain: multilingual CV/job parsing, semantic search, candidate-job matching, and skills/labor-market intelligence. Its technology is embedded inside thousands of ATS, job board, and HR platforms rather than sold direct to most end recruiters, and it claims its parsing engine is used across a large share of the global HR-tech industry. It acquired US parsing company Sovren in 2021 (whose engine and SovScore matching survive in the product) and was itself acquired by Bullhorn in 2024, now operating as 'Textkernel by Bullhorn'.
Regulatory exposure
As an Amsterdam-headquartered provider of AI used to filter and rank job applicants, Textkernel sits squarely in the EU AI Act's Annex III high-risk employment category, with provider obligations (Art. 11 technical documentation, Art. 9 risk management, Art. 14 human oversight) and downstream deployer FRIA duties (Art. 27) becoming enforceable in August 2026. Its matching engine is an Automated Employment Decision Tool under NYC Local Law 144, and its sourcing/screening output is exposed to Illinois and Colorado AI-hiring rules. Textkernel has engaged with this exposure more than most parsing vendors: it published a NYC LL 144 bias audit by Holistic AI (Feb 2025) and a 'white box' explainability stance, but offers no FRIA template, no ISO 42001, and no downloadable audit report.
Path to a higher score
The fastest gains are publishing the full Holistic AI bias-audit report (with intersectional impact-ratio tables and the AEDT candidate notice) as a durable downloadable artifact rather than a blog summary, and committing to annual re-audits to demonstrate continuity. Textkernel should produce an Article 11-style technical pack / model card and instructions-for-use per product, pursue ISO 42001 to formalize its AI management system, and ship explicit EU AI Act deployer guidance including a FRIA template - areas where it currently publishes nothing. A public trust center exposing the SOC 2 report, a data-exclusion statement, and a model-update changelog would close the remaining data-governance and post-market-monitoring gaps.
Regulatory frame
What applies to candidate sourcing AI.
Ranking outputs are inside Annex III §4 when they influence hiring decisions. GDPR Article 22 (automated decision-making) and the GDPR rights around profiling apply independently for EU candidates.
Compare
Textkernel against its nearest-scoring peers.
The closest scores in Candidate sourcing AI, side by side.
- Side-by-sideTextkernel vs Phenom52F
- Side-by-sideTextkernel vs HiredScore52F
- Side-by-sideTextkernel vs Wellfound52F
- Side-by-sideTextkernel vs Indeed50F
Ranking
Others rated in Candidate sourcing AI
Ranked 24 of 86 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.
- #21Phenom52F
- #22HiredScore52F
- #23Wellfound52F
- #24TextkernelThis profile51F
- #25Indeed50F
- #26Handshake50F
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.