Darwinbox
Darwinbox is an enterprise HCM suite whose recruitment module uses Darwinbox Sense AI (now sold inside its Cortex agent platform) to parse resumes, stack-rank and recommend candidates with stated fit reasoning, run screening and shortlisting agents, and give candidates an assistant for job matching, interview details and FAQs.
HireAIScore rates Darwinbox 46 out of 100 (grade F, Substantial gaps) under rubric v1.0, last reviewed September 25, 2026. Darwinbox ranks 41 of 77 vendors rated in ATS-integrated AI, against a category average of 47. That score is drawn from 31 evidence items across 7 rubric criteria for Darwinbox — 24 with a cited source, 7 recording that nothing was located.
§ 01 - Company facts
- Legal name
- Darwinbox Digital Solutions Pvt. Ltd.
- Founded
- 2015
- Headquarters
- India · IN
- Pricing tier
- Enterprise
- Market side
- Employer-side (AEDT)
- Categories
- ATS-integratedScreeningSourcing
- Website
- darwinbox.com
These are company details, not findings. They are editable by a verified representative and carry no weight in the score, which is built only from the cited evidence below.
§ 02 - Score breakdown
§ Score breakdown
Category scoring
Weighted contribution shown to the right of each bar.
- 01
Article 11 Technical Documentation
Weight 20%60
+12.0 · category avg 51
- 02
Bias Audit Transparency
Weight 18%28
+5.0 · category avg 41
- 03
FRIA Support
Weight 15%30
+4.5 · category avg 32
- 04
Data Governance Disclosure
Weight 15%58
+8.7 · category avg 54
- 05
Human Oversight Design
Weight 12%55
+6.6 · category avg 57
- 06
Post-Market Monitoring
Weight 12%35
+4.2 · category avg 42
- 07
Customer Documentation
Weight 8%57
+4.6 · category avg 59
Category avg is the mean raw score on that criterion across the 77 ATS-integrated AI vendors in scope of this rubric, this one included.
§ 03 - Strongest · weakest
Strongest category
Article 11 Technical Documentation
Raw score 60 · contributes 12.0 to total.
60 against a 51 category average
Weakest category
Bias Audit Transparency
Raw score 28 · contributes 5.0 to total.
28 against a 41 category average
§ 04 - Cited evidence
Download diligence record→§ Evidence
Cited per category
Every score is backed by at least one cited piece of evidence.
Evidence ledger
- Items
- 31
- Documentation
- 14
- Public statement
- 10
- Absence
- 7
- With a source URL
- 24 of 31
- Source hostnames
- 3
- Fewest items
- 2
- FRIA Support
These figures measure how thoroughly Darwinbox was reviewed, not how Darwinbox performed — an absence row, recording that nothing was located, is counted like any other item.
Article 11 Technical Documentation
5 items60
- DocumentationCaptured Sep 25, 2026
Darwinbox's self-hosted trust center lists an ISO/IEC 42001:2023 AI management system certification with TUV SUD as auditor, but the certificate is in the private, request-only tier, so whether its scope covers the Sense/Cortex recruiting features could not be checked.
- DocumentationCaptured Sep 25, 2026
The public 'Darwinbox AI Accelerator Pack' help-center PDF (February 2026) gives one-line descriptions of the recruiting AI features (resume parsing, stack ranking of applicants from most to least relevant, CV insights on relevance, job hopping and missing skills, CV-based job recommendations) and of the Sense guardrails, but it is a credit-billing document, not a technical file.
https://help.darwinbox.com/api/khub/documents/805sxOoz5cn4t6K27_p0tw/content
- Public statementCaptured Sep 25, 2026
The AI in Recruitment page's 'Innovating Responsibly' section says every AI feature is governed by a cross-functional Responsible AI Council with human oversight, risk reviews and external validation, and that every AI decision is documented and auditable.
https://darwinbox.com/innovations/artificial-intelligence-in-hr/ai-in-recruitment
- Public statementCaptured Sep 25, 2026
A March 2024 company blog says Darwinbox Sense is built on an in-house LLM called PROSE trained with sampled, anonymised data, and that Darwinbox makes no calls to public LLMs such as ChatGPT.
- AbsenceCaptured Sep 25, 2026
No model card, system card, explainability statement for the stack-ranking or fit-score logic, instructions for use, or downloadable AI policy was found on darwinbox.com, in the trust center's 16-document list, in the public help center (5 public documents), or in the newsroom, blog and explore sitemaps.
Bias Audit Transparency
4 items28
- Public statementCaptured Sep 25, 2026
Darwinbox says its models are continuously tested for bias and trained on diverse data, and that its AI hiring systems are tested for bias using structured metrics, but it names no auditor, method, date or results.
https://darwinbox.com/innovations/artificial-intelligence-in-hr/ai-in-recruitment
- Public statementCaptured Sep 25, 2026
A November 2023 company blog says the job-description generator's prompt is engineered to avoid gender, race and ethnicity bias by default, which covers job ad text, not candidate ranking.
- Public statementCaptured Sep 25, 2026
A February 2024 company blog says the resume-parsing NER model was re-engineered to interpret resumes from different global regions to improve inclusivity, without publishing any test results.
- AbsenceCaptured Sep 25, 2026
No NYC Local Law 144 bias audit summary or independent fairness audit of stack ranking, CV insights or the screening agents was found: none of the trust center's 16 documents is a bias audit, trust.warden-ai.com/darwinbox returns 404, and Darwinbox's own NYC compliance guide does not mention the law.
FRIA Support
2 items30
- DocumentationCaptured Sep 25, 2026
The trust center offers a DPA, a GDPR attestation and ISO 27701 and ISO 42001 certificates on request, which a deployer could draw on for a DPIA, but none of its 16 documents addresses the EU AI Act.
- AbsenceCaptured Sep 25, 2026
No EU AI Act statement, Article 27 FRIA template, Article 26 deployer-obligation guidance or instructions for use was found on darwinbox.com, in the trust center document list, in the public help center, or in the blog, newsroom and explore sitemaps, none of which has an AI Act page.
Data Governance Disclosure
6 items58
- DocumentationCaptured Sep 25, 2026
The trust center lists ISO 27001:2022, 27701, 27017 and 27018 (BSI), SOC 1 and SOC 2 Type II (KPMG), an EY GDPR assurance report and ISO/IEC 42001 (TUV), and describes AES-256 encryption, tenant isolation, PII masking, secure deletion and hosting regions in India, Singapore, the US, Frankfurt, Jakarta and the UAE.
- Public statementCaptured Sep 25, 2026
The candidate-data FAQ says data is isolated per tenant and that organisations can opt out of their data being used for model training, which implies customer data feeds model training unless the customer opts out.
https://darwinbox.com/innovations/artificial-intelligence-in-hr/ai-in-recruitment
- Public statementCaptured Sep 25, 2026
Darwinbox says access to historical client data lets its AI learn from past outcomes, and that its training pipelines use strategic random sampling and anonymisation.
- DocumentationCaptured Sep 25, 2026
The privacy policy says Darwinbox acts as a GDPR processor and CPRA service provider for its customers' employees and prospective employees, with the employer as controller.
- DocumentationCaptured Sep 25, 2026
The sub-processors page lists AWS, Microsoft, SendGrid, MongoDB Atlas and Office 365 and names no AI model provider, while warning that 'Innovation Service' sub-processors may not currently meet all of Darwinbox's listed measures.
- AbsenceCaptured Sep 25, 2026
No list of the input features or excluded protected attributes behind stack ranking or CV insights (which flag 'job hopping') was found on the product pages, the trust center, the help-center PDFs or the blog.
Human Oversight Design
5 items55
- Public statementCaptured Sep 25, 2026
The AI in Recruitment page says screening agents evaluate candidates semantically, explain their rankings and recommend who to interview, and that human review remains part of the process for high-impact decisions.
https://darwinbox.com/innovations/artificial-intelligence-in-hr/ai-in-recruitment
- DocumentationCaptured Sep 25, 2026
The Cortex page lists Screening, Shortlisting, Recruiter Interview and Assessment agents and says every agent step is gated, reversible, attributed and leaves a receipt.
- Public statementCaptured Sep 25, 2026
The 4 August 2026 Cortex launch release says every agent action is transparent, policy-aware and accountable, leaves a receipt of every decision, and runs within the customer's existing permissions and governance framework.
https://newsroom.darwinbox.com/darwinbox-launches-cortex-ai-native-hcm-platform
- Public statementCaptured Sep 25, 2026
Darwinbox's May 2026 listicle says Sense's AI-driven fit scoring includes explainable reasoning, so that fit scores come with the why and not just a number.
https://darwinbox.com/blog/10-best-ai-recruiting-platforms-software
- AbsenceCaptured Sep 25, 2026
No public documentation shows how a recruiter overrides a ranking or fit score, whether rejection emails or auto-rejection can run without human approval, or whether AI features can be switched off per jurisdiction such as NYC, and the help center's recruitment documentation is not publicly accessible.
Post-Market Monitoring
4 items35
- DocumentationCaptured Sep 25, 2026
The trust center describes Prisma Cloud threat monitoring with PagerDuty alerting, logged and tracked security incidents, and semi-annual penetration tests by CERT-In empanelled firms.
- DocumentationCaptured Sep 25, 2026
An Information Security and Privacy Incident Management policy is listed in the trust center but sits in the private, request-only tier.
- DocumentationCaptured Sep 25, 2026
The only public contacts found are privacy@darwinbox.com for data subject requests and dpo@darwinbox.com for grievances, with no security-specific contact anywhere on the site.
- AbsenceCaptured Sep 25, 2026
There is no public status page: status.darwinbox.com behaves like any unregistered subdomain, a tenant login that bounces to the marketing homepage; darwinbox.statuspage.io redirects to Statuspage's own site; and the security.txt and responsible-disclosure paths return 404, as does any AI incident channel or public changelog.
Customer Documentation
5 items57
- DocumentationCaptured Sep 25, 2026
The AI in Recruitment product page includes FAQs on how Darwinbox approaches fairness, human review and candidate-data security in AI hiring.
https://darwinbox.com/innovations/artificial-intelligence-in-hr/ai-in-recruitment
- DocumentationCaptured Sep 25, 2026
A public help-center PDF sets out each recruiting AI feature, the candidate and onboarding agents, and how Sense Credits are charged.
https://help.darwinbox.com/api/khub/documents/805sxOoz5cn4t6K27_p0tw/content
- DocumentationCaptured Sep 25, 2026
The trust center catalogues 16 documents, including a DPA, SOC reports, ISO certificates and policies, but only the privacy policy is public; the rest are released on request under NDA.
- DocumentationCaptured Sep 25, 2026
Darwinbox's January 2026 guide to NYC HR compliance covers pay transparency and leave laws but never mentions Local Law 144, automated employment decision tools or bias audits.
https://darwinbox.com/blog/nyc-hr-compliance-local-mandates-2025
- AbsenceCaptured Sep 25, 2026
No deployer guidance was found for the EU AI Act, NYC LL 144, Illinois HB 3773 or the Colorado and California automated-decision rules, and no candidate-notice templates, across the product pages, trust center, public help center, blog, newsroom and explore sitemaps.
§ 05 - Editorial notes
Company overview
Darwinbox (Darwinbox Digital Solutions Pvt. Ltd.) is an enterprise HCM vendor founded in 2015 and headquartered in Hyderabad, India. At its August 2026 Cortex launch it reported more than 1,400 enterprise customers and 4.5 million employees on the platform, with backers including KKR, Partners Group, Microsoft and Salesforce Ventures. Recruitment is one module of the suite. Its AI layer, Darwinbox Sense, is now marketed inside the Cortex agent platform, and the old Sense page redirects to Cortex. Sense parses resumes, stack-ranks applicants from most to least relevant, and shows 'CV insights' such as relevance, job hopping and missing skills. It also recommends candidates from the talent pool and runs Screening, Shortlisting, Recruiter Interview and Assessment agents. Candidates use a Candidate Agent for CV-based job recommendations, interview details and company FAQs, and the product page promises rejection emails with 'contextual feedback'. AI use is billed in Sense Credits on top of per-employee-per-month pricing.
Regulatory exposure
Stack ranking, fit scoring and the screening and shortlisting agents rank and filter applicants. NYC employers that rely on them are therefore likely using an AEDT under Local Law 144, which requires an annual independent bias audit and candidate notice. Darwinbox publishes no audit, and its January 2026 NYC compliance guide never mentions the law. DCWP has promised tougher enforcement after the December 2025 State Comptroller audit. In the EU, the same features are high-risk under Annex III point 4, which makes Darwinbox a provider (it offers Frankfurt hosting). The Digital AI Omnibus deferred those obligations to 2 December 2027. The ISO 42001 certification listed in its trust center is a head start, but Darwinbox has published no technical documentation, instructions for use or AI Act statement. No emotion-recognition feature was found. Under HB 3773, in force since 1 January 2026, Illinois employers must give notice of AI use and should test signals such as 'job hopping' for proxy discrimination. Colorado's SB 26-189 adds disclosure duties from 1 January 2027. In California, the FEHA automated-decision rules have applied since 1 October 2025, including four-year record retention, and CPPA ADMT rules follow on 1 January 2027.
Path to a higher score
The biggest gains are in bias transparency and deployer support. Darwinbox should commission and publish an independent Local Law 144-style bias audit of stack ranking, fit scoring and the screening agents, naming the auditor, date and impact ratios, and repeat it every year. It should publish the scope of its TUV SUD ISO 42001 certificate and a model card for the ranking logic. The card should list inputs, excluded attributes (including how job hopping and tenure gaps are treated) and validation results. It should also publish instructions for use and EU AI Act deployer guidance with FRIA and candidate-notice templates. On oversight, it should document how recruiters override rankings, whether a human approves AI rejection emails, and which features can be switched off per jurisdiction. Operationally, it needs a public status page, a security.txt or vulnerability-disclosure page, public release notes and named AI sub-processors. Using customer data for model training should be opt-in rather than opt-out.
§ Regulatory frame
What applies to ats-integrated ai.
Ranking features are inside EU AI Act Annex III §4 (high-risk) and inside NYC Local Law 144 when they affect NYC-based candidates. Customers should treat the AI surface as a distinct deployer-FRIA scope from the rest of the ATS.
§ Compare
Build any comparison→Darwinbox against its nearest-scoring peers.
In ATS-integrated AI.
§ Others rated in ATS-integrated AI
All ats-integrated vendors→Ranked 41 of 77 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.
- 38Factorial46F
- 39ClearCo (ClearCompany)46F
- 40Talkpush46F
- 41DarwinboxThis profile46F
- 42JobAdder45F
- 43Jobvite45F
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.