# Duomian ·AI Mianshi

The company’s AI product is Duomian AI Interview.
1. Performance
Built independently using McClelland’s iceberg and competency models, Liepin’s years of HR expertise, practical experience and specialist data, plus AI. Emphasizes accurate assessment, ease and efficiency, and value.
2. Applications
Serves large companies and SMEs in campus and experienced hiring, assessing university students and white-, grey- and blue-collar workers across recruitment scenarios:
1) Hiring: consistent scientific role criteria and rapid AI screening.
2) Probation: assess whether employees meet permanent-role requirements.
3) Talent reviews: assess development potential to optimize staffing.
4) Transfers and promotions: role-specific models and questions assess fit and identify stronger candidates.
3. Technical innovation
1) Validity and reliability: LLMs and tuned algorithms achieve over 90% scoring consistency with senior interviewers.
2) Broad iceberg-model dimensions: 7 levels and nearly a hundred dimensions match people, roles and organizations.
3) Evidence-Based (EB) technology: claimed industry-first multi-layer probing improves assessment precision.
4) Resume-based personalized questions.
5) Comprehensive intelligent anti-cheating for fairness.
6) Multimodal analysis of voice, semantics and images.
7) Multilingual scoring with fluent Cantonese and English and extensibility to other languages/dialects.
8) Custom rooms, interviewers and questions reinforce employer identity.
4. Core advantages and industry needs
1) Accurate assessment: 1 model, 3 questioning methods, 2 evaluations and 1 anti-cheating approach.
A. 1 model: scientific and intelligent modeling.
Covers 7 iceberg-model levels and 140+ dimensions, including physical qualities, knowledge/skills, role experience, ability, character, motivation and values. Recruiters enter a role and basic requirements; Doris generates an editable JD, then a competency model based on the confirmed requirements, iceberg theory and industry data, with manual adjustment supported.
B. 3 questioning methods: competency questions, proprietary resume-analysis questions and pioneering EB deep probing.
Generates editable interview questions from competency models and recruiter needs. Automatically identifies resume strengths and risks for targeted, realistic individual interviews. Patented Evidence-Based multi-stage follow-ups aim to reveal actual capabilities.
C. 2 evaluations: anchor-comparison scoring and multimodal assessment.
The team argues LLMs cannot directly evaluate candidates reliably. Its proprietary anchor-comparison method combines models and industry expertise, scoring against key dimension anchors for more scientific precision. Voice, meaning, expressions and body movement inform holistic assessment, replicating human interview practice.
D. 1 anti-cheating approach throughout.
The source reports candidate cheating above 15%, with conventional methods insufficient. LLM-assisted answer analysis identifies cheating using large models, described as an industry first.
2) Efficient and easy to use
Create an interview in four steps:
a. Confirm role information and resume-question elements.
b. Generate the competency model from the role.
c. Set interviewer appearance, background, language and opening remarks.
d. Configure interview behavior requirements.
Customize questions, rooms and interviewers:
a. Tailor dimensions, scoring standards and questions to enterprise competency models.
b. Match employer branding through rooms, opening statements, automated answers, promotional videos and warm-up questions.
c. Customize 2D or 3D digital interviewers to improve candidate engagement and employer presentation.
Rich interview reports and filterable report lists help evidence-based screening. Smooth experiences across desktop, mini programs and apps fit different scenarios.
Multilingual interviews support local talent needs. Speech recognition and NLP enable natural English, Cantonese and other language/dialect interactions, letting candidates express themselves fully and contributing diverse perspectives to inclusive talent ecosystems.
3) Value: save money, time and effort
Money: interview costs are 1/5-1/8 of offline interviews; attendance rises from 60% to 85%.
Time: average interview execution cycles fall to 60%~70% of their original duration.
Effort: intelligent resume questions identify strengths/risks, patented EB follow-ups probe deeply, anchor-based scoring improves precision and multimodal evaluations combine meaning, voice and expressions. LLM-based anti-cheating addresses impersonation, leaked questions, looking up information and memorized answers, with claimed detection accuracy of 95%.

The company’s AI product is Duomian AI Interview.
1. Performance
Built independently using McClelland’s iceberg and competency models, Liepin’s years of HR expertise, practical experience and specialist data, plus AI. Emphasizes accurate assessment, ease and efficiency, and value.
2. Applications
Serves large companies and SMEs in campus and experienced hiring, assessing university students and white-, grey- and blue-collar workers across recruitment scenarios:
1) Hiring: consistent scientific role criteria and rapid AI screening.
2) Probation: assess whether employees meet permanent-role requirements.
3) Talent reviews: assess development potential to optimize staffing.
4) Transfers and promotions: role-specific models and questions assess fit and identify stronger candidates.
3. Technical innovation
1) Validity and reliability: LLMs and tuned algorithms achieve over 90% scoring consistency with senior interviewers.
2) Broad iceberg-model dimensions: 7 levels and nearly a hundred dimensions match people, roles and organizations.
3) Evidence-Based (EB) technology: claimed industry-first multi-layer probing improves assessment precision.
4) Resume-based personalized questions.
5) Comprehensive intelligent anti-cheating for fairness.
6) Multimodal analysis of voice, semantics and images.
7) Multilingual scoring with fluent Cantonese and English and extensibility to other languages/dialects.
8) Custom rooms, interviewers and questions reinforce employer identity.
4. Core advantages and industry needs
1) Accurate assessment: 1 model, 3 questioning methods, 2 evaluations and 1 anti-cheating approach.
A. 1 model: scientific and intelligent modeling.
Covers 7 iceberg-model levels and 140+ dimensions, including physical qualities, knowledge/skills, role experience, ability, character, motivation and values. Recruiters enter a role and basic requirements; Doris generates an editable JD, then a competency model based on the confirmed requirements, iceberg theory and industry data, with manual adjustment supported.
B. 3 questioning methods: competency questions, proprietary resume-analysis questions and pioneering EB deep probing.
Generates editable interview questions from competency models and recruiter needs. Automatically identifies resume strengths and risks for targeted, realistic individual interviews. Patented Evidence-Based multi-stage follow-ups aim to reveal actual capabilities.
C. 2 evaluations: anchor-comparison scoring and multimodal assessment.
The team argues LLMs cannot directly evaluate candidates reliably. Its proprietary anchor-comparison method combines models and industry expertise, scoring against key dimension anchors for more scientific precision. Voice, meaning, expressions and body movement inform holistic assessment, replicating human interview practice.
D. 1 anti-cheating approach throughout.
The source reports candidate cheating above 15%, with conventional methods insufficient. LLM-assisted answer analysis identifies cheating using large models, described as an industry first.
2) Efficient and easy to use
Create an interview in four steps:
a. Confirm role information and resume-question elements.
b. Generate the competency model from the role.
c. Set interviewer appearance, background, language and opening remarks.
d. Configure interview behavior requirements.
Customize questions, rooms and interviewers:
a. Tailor dimensions, scoring standards and questions to enterprise competency models.
b. Match employer branding through rooms, opening statements, automated answers, promotional videos and warm-up questions.
c. Customize 2D or 3D digital interviewers to improve candidate engagement and employer presentation.
Rich interview reports and filterable report lists help evidence-based screening. Smooth experiences across desktop, mini programs and apps fit different scenarios.
Multilingual interviews support local talent needs. Speech recognition and NLP enable natural English, Cantonese and other language/dialect interactions, letting candidates express themselves fully and contributing diverse perspectives to inclusive talent ecosystems.
3) Value: save money, time and effort
Money: interview costs are 1/5-1/8 of offline interviews; attendance rises from 60% to 85%.
Time: average interview execution cycles fall to 60%~70% of their original duration.
Effort: intelligent resume questions identify strengths/risks, patented EB follow-ups probe deeply, anchor-based scoring improves precision and multimodal evaluations combine meaning, voice and expressions. LLM-based anti-cheating addresses impersonation, leaked questions, looking up information and memorized answers, with claimed detection accuracy of 95%.

Canonical: https://uniquebase.ai/products/duomian-ai-mianshi
Stable identifier: UB-D916
Snapshot captured: 2026-10-20T18:06:34.551Z
English updated: 2026-10-20T18:30:14.490Z

## Connected profiles

- [Duomian](/companies/duomiankeji)

## Sources

- No source links are available for this profile.