Emi
HireAIScore rates Emi 33 out of 100 (grade F, Substantial gaps) under rubric v1.0, last reviewed September 25, 2026. Emi ranks 153 of 160 vendors rated in Screening AI, against a category average of 46. That score is drawn from 26 evidence items across 7 rubric criteria for Emi — 19 with a cited source, 7 recording that nothing was located.
Score breakdown
Emi, 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 49
28
+5.6
Bias Audit Transparency
Weight 18% · category avg 41
18
+3.2
FRIA Support
Weight 15% · category avg 31
25
+3.8
Data Governance Disclosure
Weight 15% · category avg 53
44
+6.6
Human Oversight Design
Weight 12% · category avg 56
44
+5.3
Post-Market Monitoring
Weight 12% · category avg 40
35
+4.2
Customer Documentation
Weight 8% · category avg 58
48
+3.8
Total weighted score
32.5(rounded to 33 for display)
Category avg is the mean raw score on that criterion across the 160 Screening AI vendors in scope of this rubric, this one included.
Strongest criterion
Customer Documentation
48raw / 100
Contributes 3.8 to the total. 48 against a 58 category average.
Weakest criterion
Bias Audit Transparency
18raw / 100
Contributes 3.2 to the total. 18 against a 41 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
- 11
- Public statement
- 8
- Absence
- 7
- With a source URL
- 19 of 26
- Source hostnames
- 3
- Fewest items
- 2
- FRIA Support
These figures measure how thoroughly Emi was reviewed, not how Emi performed — an absence row, recording that nothing was located, is counted like any other item.
Article 11 Technical Documentation
4 items28
- DocumentationCaptured Sep 25, 2026
The product page describes screening only in marketing terms ('Let Emi's AI evaluate and filter candidates automatically with intelligent scoring') and shows example fit percentages without explaining inputs, weighting or limits, though it details the Digital File agent's roughly 50 document cross-checks against SAT, IMSS and RENAPO.
- Public statementCaptured Sep 25, 2026
The only public explanation of how filtering works is an August 2023 blog post, which says Emi applies rules set by the hiring team and that this automation does not use continuous, autonomous learning the way AI does.
https://www.emilabs.ai/es/blog/chatbots-de-inteligencia-artificial
- Public statementCaptured Sep 25, 2026
The current home page says '100% of the conversation runs on generative AI, no decision trees', so the 2023 rule-based description no longer covers the product, and no updated technical description has been published.
- AbsenceCaptured Sep 25, 2026
No AI policy, model or system card, explainability statement for the fit percentage, instructions for use, ISO/IEC 42001 certificate or EU AI Act documentation turned up anywhere in emilabs.ai's sitemap (product, retention, security, about and integrations pages, 11 client stories, 40 Spanish blog posts and 10 guides) or in web searches, and help.emilabs.ai redirects to a customer login.
Bias Audit Transparency
3 items18
- AbsenceCaptured Sep 25, 2026
No NYC LL 144 bias audit, independent audit (BABL AI, DCI, ORCAA, Warden AI, Holistic AI, Credo AI, ConductorAI) or academic audit of Emi turned up on emilabs.ai or in web searches, and Warden AI's dashboard returns 404 for the emi, emilabs and emi-labs slugs.
- Public statementCaptured Sep 25, 2026
An October 2023 post says pre-screening runs on 'customized evaluation rules provided by employers' and that Emi's automation 'can eliminate the subjectivity' of manual evaluation, a fairness claim with no published testing behind it.
- Public statementCaptured Sep 25, 2026
A March 2023 best-practice post tells customers to keep questions about age, race or gender out of chatbot flows because they could lead to discrimination, which is design advice rather than evidence that the fit score has been tested for adverse impact.
FRIA Support
2 items25
- AbsenceCaptured Sep 25, 2026
No Fundamental Rights Impact Assessment template, EU AI Act deployer guidance or high-risk-system material turned up on emilabs.ai, its Spanish resources hub or in web searches, and help.emilabs.ai is behind a login.
- DocumentationCaptured Sep 25, 2026
The privacy policy lists GDPR data-subject rights and standard contractual clauses for transfers to the US, Argentina and Mexico, which is standard data-protection text rather than support for an Article 27 assessment.
Data Governance Disclosure
6 items44
- DocumentationCaptured Sep 25, 2026
The security page claims SOC 2 Type II certification, SSO, role and permission management and multi-cloud hosting, but gives no detail on encryption, hosting region, sub-processors or retention, and offers no way to request the report.
- Public statementCaptured Sep 25, 2026
An August 2023 post announces Emi's first SOC 2 Type 2 report, audited by Insight Assurance and covering vendor risk, IT infrastructure, secure development, change management, network and access security, physical and endpoint security, and identity lifecycle management.
https://www.emilabs.ai/es/blog/emi-compromiso-con-seguridad-soc2
- Public statementCaptured Sep 25, 2026
A December 2023 post says Emi again passed a SOC 2 Type II examination by Insight Assurance, adding the Availability criterion, and no later report date has been published even though the site still shows a SOC 2 Type II badge.
- DocumentationCaptured Sep 25, 2026
The privacy policy (updated 2 May 2024) says Emi collects the ID documents, education and work history candidates share over WhatsApp and keeps candidate data under 'retention policies agreed upon with each employer client', but it names no sub-processors or retention periods and says nothing about model training or automated decisions.
- DocumentationCaptured Sep 25, 2026
The Digital File agent extracts CURP, RFC, NSS, address and bank details from documents sent over WhatsApp and checks them live against SAT, IMSS and RENAPO, yet Emi publishes no retention or deletion rule for these documents.
- AbsenceCaptured Sep 25, 2026
No trust center, sub-processor list, training-data disclosure or data-exclusion list turned up: trust.emilabs.ai and emilabs.safebase.io do not resolve, and app.vanta.com/emilabs/trust redirects to Vanta's sign-in page in a real browser.
Human Oversight Design
4 items44
- DocumentationCaptured Sep 25, 2026
The product page says Emi's AI will 'evaluate and filter candidates automatically' so that 'your team only receives candidates who meet the role requirements' and shows per-candidate fit percentages with no documented explanation or override, while document checks use green, amber and red statuses so staff review 'only what needs a human eye'.
- Public statementCaptured Sep 25, 2026
The August 2023 post presents Emi as a tool for making informed decisions that does not replace recruiters, and says candidates who fail the hiring team's rules go to a separate list that recruiters can re-check manually.
https://www.emilabs.ai/es/blog/chatbots-de-inteligencia-artificial
- Public statementCaptured Sep 25, 2026
Heineken México's privacy notice for candidates who apply through Emilabs says their data may be subject to automated decision-making 'without the intervention of a natural person's assessment' in recruitment and selection, so at least one customer tells applicants that decisions may be fully automated.
https://heinekenmexico.com/aviso-de-privacidad-candidatos-emilabs
- AbsenceCaptured Sep 25, 2026
No public documentation of reason codes or explanations for the fit score, audit logs, mandatory review of filtered-out candidates or per-jurisdiction settings turned up on emilabs.ai, and the help portal is behind a login.
Post-Market Monitoring
3 items35
- DocumentationCaptured Sep 25, 2026
An Atlassian Statuspage tracks 11 components in three groups, including a Chatbot group covering WhatsApp, SMS and Facebook plus the app.emilabs.ai and my.emilabs.ai front and back ends, but it is not linked from emilabs.ai and status.emilabs.ai does not resolve.
- DocumentationCaptured Sep 25, 2026
The incident log holds three incidents (a June 2023 login outage, intermittent web-platform problems in August 2023 and a January 2025 backoffice slowdown), two of them with root-cause notes, and nothing has been logged since January 2025.
- AbsenceCaptured Sep 25, 2026
No security contact, vulnerability-disclosure policy, security.txt (emilabs.ai/.well-known/security.txt returns 404), AI model changelog or fairness-monitoring page turned up, and the only contact the site publishes is privacy@emilabs.ai.
Customer Documentation
4 items48
- DocumentationCaptured Sep 25, 2026
Public product pages cover the attract, screen, schedule, Digital File and analytics agents, with separate pages for retention, integrations and security.
- DocumentationCaptured Sep 25, 2026
Customer stories such as Danone's in Mexico City describe deployments, including 'Emi's AI candidate scoring' and automatic assignment of candidates to the workplace closest to their residence.
- DocumentationCaptured Sep 25, 2026
The Spanish resources hub lists blog posts and guides on frontline recruiting and Mexican labour topics, none of them on AI compliance, bias or deployer duties.
- AbsenceCaptured Sep 25, 2026
No public terms of service, DPA, sub-processor list or deployer guidance for NYC, the EU or US states turned up (/en/terms, /en/dpa and /en/subprocessors return 404), and help.emilabs.ai redirects to a customer login on app.emilabs.ai.
Editorial notes
Company overview
Emi (Emi Labs, Inc.) was founded in Buenos Aires in 2017 by Mateo Cavasotto and Andrés Arslanian. It went through Y Combinator's Winter 2019 batch and raised an $11M Series A in 2022, led by Merus Capital and Khosla Ventures. The parent company gives a San Francisco address and works through subsidiaries that include Emi Labs Mexicana, while its engineering team is in Buenos Aires. Emi sells AI agents for high-volume frontline hiring, mostly in Mexico, where it says 1 in 10 frontline hires go through its platform. The agents chat with candidates on WhatsApp, SMS and Messenger. They score and filter candidates against employer rules, giving each a fit percentage, and book interviews. They also collect ID, tax and social-security documents and check them against Mexico's RENAPO, IMSS and SAT databases. A newer retention product interviews former staff and checks in monthly with current employees. Every published case study is a Mexican operation, including OXXO, Home Depot México, Heineken México and Danone.
Regulatory exposure
Because it scores and filters job applicants, Emi's screening agent falls in the EU AI Act's high-risk recruitment category (Annex III point 4). The retention agent's monthly monitoring of current employees for risk signals could fall under the same point. The Digital AI Omnibus pushed those obligations back to 2 December 2027, and no EU deployments were found. If an employer used the fit score for New York City roles, it would likely count as an automated employment decision tool under Local Law 144, which requires an annual independent bias audit and candidate notice. Emi has published no audit. Its US exposure looks limited: all its case studies are Mexican, though TechCrunch reported US customers in 2022 and the privacy policy lists California rights. US customers would face three sets of rules. Illinois HB 3773 requires notice and bans zip codes as proxies, which matters because Emi routes candidates to the workplace nearest their home. California's FEHA automated-decision rules require four-year record retention, and Emi leaves retention to each customer's contract. Colorado's SB 26-189 disclosure regime starts on 1 January 2027.
Path to a higher score
Emi's weakest areas are bias testing and technical documentation. An independent bias audit of the fit score and automatic filter would do the most for its score, reporting impact ratios by sex and race or ethnicity and repeated every year. A published system description should replace the 2023 rule-based account. It should explain what the fit percentage measures, how employer rules and the generative-AI agents combine, and where they fall short. Emi should document its oversight controls: whether filtered-out candidates always land in a list someone can review, what reason recruiters see, override and audit-log features, and guidance against fully automated rejections. For data governance, it should name its sub-processors, including its WhatsApp and model providers. It should also publish retention periods for chat transcripts and for the ID, tax, social-security and bank documents it collects, its model-training policy, and the date of its latest SOC 2 report. Linking the status page and publishing a security contact, a DPA and deployer guidance for NYC, the EU and US states would lift the remaining scores.
Regulatory frame
What applies to screening AI.
Treated as high-risk under Annex III §4 when the screening output is decisive. EU AI Act Article 13 transparency obligations apply; deployers must give candidates a way to contest.
Compare
Emi against its nearest-scoring peers.
The closest scores in Screening AI, side by side.
- Side-by-sideEmi vs HireLogic34F
- Side-by-sideEmi vs Pulsifi33F
- Side-by-sideEmi vs Qureos33F
- Side-by-sideEmi vs Entelo33F
Ranking
Others rated in Screening AI
Ranked 153 of 160 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.
- #150Pulsifi33F
- #151Qureos33F
- #152Entelo33F
- #153EmiThis profile33F
- #154Arena Analytics32F
- #155Intervue.io31F
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.