Avionté
HireAIScore rates Avionté 45 out of 100 (grade F, Substantial gaps) under rubric v1.0, last reviewed September 25, 2026. Avionté ranks 46 of 77 vendors rated in ATS-integrated AI, against a category average of 47. That score is drawn from 33 evidence items across 7 rubric criteria for Avionté — 27 with a cited source, 6 recording that nothing was located.
Score breakdown
Avionté, 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 51
50
+10.0
Bias Audit Transparency
Weight 18% · category avg 41
22
+4.0
FRIA Support
Weight 15% · category avg 32
35
+5.3
Data Governance Disclosure
Weight 15% · category avg 54
55
+8.3
Human Oversight Design
Weight 12% · category avg 57
60
+7.2
Post-Market Monitoring
Weight 12% · category avg 42
45
+5.4
Customer Documentation
Weight 8% · category avg 59
62
+5.0
Total weighted score
45.0(rounded to 45 for display)
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.
Strongest criterion
Customer Documentation
62raw / 100
Contributes 5.0 to the total. 62 against a 59 category average.
Weakest criterion
Bias Audit Transparency
22raw / 100
Contributes 4.0 to the total. 22 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
- 33
- Documentation
- 22
- Public statement
- 5
- Absence
- 6
- With a source URL
- 27 of 33
- Source hostnames
- 7
- Fewest items
- 3
- FRIA Support
These figures measure how thoroughly Avionté was reviewed, not how Avionté performed — an absence row, recording that nothing was located, is counted like any other item.
Article 11 Technical Documentation
4 items50
- DocumentationCaptured Sep 25, 2026
The PIXEL GenAI article says Claude runs the chat but only extracts answers; fixed, recruiter-weighted formulas for each question type compute the 0-100 rating, and freeform answers are not scored because of fairness concerns about LLMs.
- DocumentationCaptured Sep 25, 2026
The Matching Agent article documents the five ranking factors (pay, location, position categories/skills, resume keyword match, company history), the data each compares, weight settings, filters and exclusions applied before ranking, and the Match Reason shown for each candidate.
https://avionteboldsupport.zendesk.com/hc/en-us/articles/50623227217555-Matching-Agent
- Public statementCaptured Sep 25, 2026
The March 2025 launch release sets out three AI principles (vetting AI vendors for privacy, security and ethics; human oversight of AI output; transparency in AI workflows) and says the PIXEL chatbot is built with Anthropic's Claude.
- AbsenceCaptured Sep 25, 2026
No model or system card, ISO 42001 certification, NIST AI RMF or other framework reference, or standalone AI policy page turned up in the avionte.com sitemap or site search (/ai-policy/, /responsible-ai/ and /ai-principles/ all return 404), in either Zendesk help center, or on info.avionte.com.
Bias Audit Transparency
5 items22
- AbsenceCaptured Sep 25, 2026
No NYC LL 144 bias-audit summary or independent audit (BABL AI, DCI, ORCAA, Warden AI, Holistic AI, Credo AI) covering the Matching Agent or PIXEL turned up in avionte.com site search, the search APIs of both Zendesk help centers, or at trust.warden-ai.com/avionte (404).
- DocumentationCaptured Sep 25, 2026
Avionté's Artificial Intelligence Addendum (updated 17 Oct 2025) makes the client solely responsible for any required bias or disparate-impact testing and disclaims any obligation to monitor or ensure the client's compliance.
- Public statementCaptured Sep 25, 2026
An August 2026 recap of Avionté's CONNECT keynote quotes the CTO saying bias testing and explainability are 'built in', with no published methodology, results or auditor.
https://www.avionte.com/blog/connect-2026-product-technology-keynote/
- DocumentationCaptured Sep 25, 2026
Avionté says it uses the LLM only to extract answers because it is concerned about the potential adverse impact of LLMs as judges, and it leaves freeform answers unscored; this is a design mitigation, not an audit.
- DocumentationCaptured Sep 25, 2026
The Enhanced EEO Report gives deployers self-reported demographics by job title for applicant-versus-hire funnel analysis, but it is not tied to Matching Agent or PIXEL outputs and is not an audit of either tool.
FRIA Support
3 items35
- DocumentationCaptured Sep 25, 2026
The Matching Agent article's 'Before You Enable This Feature' section asks deployers to get internal sign-off, notice and consent documentation (it cites NYC Local Law 144) and a named recordkeeping owner before enabling, and notes that Avionté does not define that process for them.
https://avionteboldsupport.zendesk.com/hc/en-us/articles/50623227217555-Matching-Agent
- DocumentationCaptured Sep 25, 2026
The AI Addendum warns that employment uses may require extra disclosures, fairness or bias testing, and human oversight, and makes identifying and meeting those requirements the client's job.
- AbsenceCaptured Sep 25, 2026
No EU AI Act Article 27 FRIA template or EU deployer-obligation guidance turned up on avionte.com (site search, sitemap, security and GDPR pages), in the BOLD or Classic help centers, or in the info.avionte.com addenda.
Data Governance Disclosure
7 items55
- DocumentationCaptured Sep 25, 2026
The AI Addendum commits that Avionté will not use client data to train its own models, or let third parties train theirs, without the client's prior consent, though it may use client data to improve the Software.
- DocumentationCaptured Sep 25, 2026
The privacy policy (updated 6 Feb 2026) says AI systems are covered by role-based access controls, audit logs and encryption, that PII is not shared with 'unauthorized external AI systems', and that third-party AI vendors are contractually bound.
- DocumentationCaptured Sep 25, 2026
The Security & Data Protection FAQ states annual SOC 2 Type II audits by independent third parties, AES-256 and TLS encryption, AWS hosting, deletion within 30 days of a request, 90-day post-contract retention and 72-hour breach notification.
https://www.avionte.com/glossary/faq-security-data-protection/
- Public statementCaptured Sep 25, 2026
A June 2026 Avionté post says its AI capabilities are 'powered by Anthropic and hosted on AWS infrastructure' and that the platform is SOC 2 Type II compliant.
https://www.avionte.com/blog/your-ai-is-only-as-good-as-the-platform-underneath-it/
- Public statementCaptured Sep 25, 2026
An AWS case study says Avionté began an Amazon Bedrock proof of concept using Titan Embeddings and Anthropic's Claude as recruiter copilots for matching workers to job postings.
https://aws.amazon.com/solutions/case-studies/avionte-case-study/
- DocumentationCaptured Sep 25, 2026
The PIXEL GenAI article says Claude has no access to candidate PII, interview transcripts are kept for one year, and uploaded attachments are not read for scoring.
- AbsenceCaptured Sep 25, 2026
No public subprocessor list or DPA was found: /subprocessors/, /sub-processors/ and /dpa/ on avionte.com return 404, a help-center search for 'subprocessor' returns nothing, and neither the MSA nor the addenda include a subprocessor schedule.
Human Oversight Design
6 items60
- DocumentationCaptured Sep 25, 2026
Recruiters see a plain-language Match Reason for each candidate's position (including why they are not ranked higher), can re-weight or zero out any of the five factors, and are told that final decisions must involve human review of each full profile.
https://avionteboldsupport.zendesk.com/hc/en-us/articles/50623227217555-Matching-Agent
- DocumentationCaptured Sep 25, 2026
The Summer 2026 release notes say both Matching Agent permissions default to disabled, tell customers to consult legal counsel before enabling, and state that rankings are meant to inform recruiter judgment, not replace it.
https://support.avionte.com/hc/en-us/articles/54854520807187-Summer-Release-2026
- DocumentationCaptured Sep 25, 2026
PIXEL automatically offers follow-up interview times only to candidates who pass every Required question and meet the recruiter-set passing percentage, so screened-out candidates get no interview offer without a human step.
https://avionteboldsupport.zendesk.com/hc/en-us/articles/52618471731603-PIXEL-Bots
- DocumentationCaptured Sep 25, 2026
Each PIXEL GenAI interview opens with an AI-use disclosure and consent request; a candidate who declines is recorded as 'Did Not Consent' rather than failed, and recruiters can review full transcripts.
- DocumentationCaptured Sep 25, 2026
The AI Addendum bars clients from making automated employment decisions that have not been subjected to human review and approval.
- Public statementCaptured Sep 25, 2026
The September 2026 Feature Spotlight says the Matching Agent is 'enabled by default and requires no setup', which appears to conflict with the help center's statement that its permissions default to disabled.
https://www.avionte.com/blog/feature-spotlight-ai-matching-agent/
Post-Market Monitoring
4 items45
- DocumentationCaptured Sep 25, 2026
status.avionte.com is a public Statuspage whose components include Avionté PIXEL, Resume Parsing and Talent Search, with dated incident histories (three PIXEL incidents in 2026) and email, SMS, webhook and RSS subscriptions, but no postmortems.
- DocumentationCaptured Sep 25, 2026
Public release notes in the support center record AI feature changes, for example the AI Matching Agent replacing the keyword-based auto-matcher on 27 Aug 2026.
https://support.avionte.com/hc/en-us/articles/54854520807187-Summer-Release-2026
- DocumentationCaptured Sep 25, 2026
The Security FAQ describes continuous vulnerability scanning, routine penetration tests and a commitment to notify customers of a breach within 72 hours.
https://www.avionte.com/glossary/faq-security-data-protection/
- AbsenceCaptured Sep 25, 2026
No security.txt (/.well-known/security.txt returns 404), vulnerability-disclosure program, dedicated security contact, AI incident-reporting channel, or model-performance or fairness monitoring dashboard was found.
Customer Documentation
4 items62
- DocumentationCaptured Sep 25, 2026
The AI & Automation product page lists the AI Matching Agent, Avionté PIXEL screening, and AI generators for job descriptions and interview scripts, and links to the privacy policy, security FAQ and status page.
- DocumentationCaptured Sep 25, 2026
The public Matching Agent article gives deployers an AI employment-law notice, cites NYC Local Law 144 disclosures, and warns against E-Verify misuse and about disparate impact from heavy location or company-history weighting.
https://avionteboldsupport.zendesk.com/hc/en-us/articles/50623227217555-Matching-Agent
- DocumentationCaptured Sep 25, 2026
Avionté publishes its Master Services Agreement (updated 17 Oct 2025), which links to a public addenda page containing the Artificial Intelligence Addendum.
- AbsenceCaptured Sep 25, 2026
No published DPA and no standalone guide to NYC LL 144, Illinois HB 3773, Colorado SB 26-189, California's automated-decision rules or the EU AI Act was found on avionte.com or in the help centers.
Editorial notes
Company overview
Avionté (Avionté, LLC) was founded in 2005 and is based in Bloomington, Minnesota. It sells AviontéBOLD, a front- and back-office platform for staffing agencies, and an AWS case study says it serves more than 1,800 agencies across North America. It has two hiring-AI features. The Matching Agent, live since 27 August 2026, ranks up to 25 candidates from an agency's database for each job. It weighs five factors the recruiter can adjust (pay, location, skills, resume keywords and company history) and explains each ranking in plain language. PIXEL is a text and app pre-screening chatbot built on Anthropic's Claude since 2025. Claude only pulls answers out of the conversation; rules the recruiter sets turn them into the score that decides whether a candidate is offered an interview. Avionté says its AI is 'powered by Anthropic and hosted on AWS infrastructure.'
Regulatory exposure
NYC Local Law 144 covers employment agencies, not just employers. A staffing firm that ranks candidates with the Matching Agent for New York City roles, or uses PIXEL scores to decide who gets an interview, may be using an automated employment decision tool. That would require an annual independent bias audit and candidate notice. Avionté publishes no audit, and its AI Addendum makes clients solely responsible for bias and disparate-impact testing. DCWP has promised tougher enforcement after the December 2025 Comptroller audit. Illinois HB 3773 (in force since 1 January 2026) requires notice of AI use and bars zip-code proxies. That makes the Matching Agent's location factor sensitive: Avionté's own documentation warns it can correlate with race or national origin. PIXEL opens every interview with an AI-use disclosure, which helps with notice duties such as Colorado SB 26-189 (from 1 January 2027). But PIXEL keeps transcripts for only one year, while California's FEHA automated-decision rules require four years. Avionté's customers are in North America, so EU AI Act exposure looks limited; the Annex III duties now start on 2 December 2027.
Path to a higher score
The biggest gain would be an independent bias audit of the Matching Agent and PIXEL scoring, in the form LL 144 requires, with the summary published and repeated every year. Next, Avionté should publish a subprocessor list that names Anthropic and AWS, a downloadable DPA, and the standalone AI policy its CTO says is already published. It should also fix conflicting documentation: one source says the Matching Agent is on by default and another says it is off, and two articles give different rules for PIXEL's commute scoring. A model card for the Matching Agent should cover its inputs, how Match Reasons are generated and how it was validated. Deployers need notice templates for each jurisdiction, an alternative-process option for candidates, and retention settings long enough for California. Finally, a vulnerability-disclosure contact and postmortems for incidents that hit AI features would round out monitoring.
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
Avionté against its nearest-scoring peers.
The closest scores in ATS-integrated AI, side by side.
- Side-by-sideAvionté vs JobAdder45F
- Side-by-sideAvionté vs Jobvite45F
- Side-by-sideAvionté vs Clevry45F
- Side-by-sideAvionté vs Dover45F
Ranking
Others rated in ATS-integrated AI
Ranked 46 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.
- #43Jobvite45F
- #44Clevry45F
- #45Dover45F
- #46AviontéThis profile45F
- #47Cegid45F
- #48Vincere44F
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.