Comsalo

Research

We publish our methodology, accuracy benchmarks, and bias audits openly. AI that affects people's careers deserves scrutiny — and we invite it.

Bias & Fairness

We measure and eliminate demographic bias across gender, age, name origin, and language. Quarterly audit results are published here.

Accuracy

We benchmark our scoring models against senior HR professionals to quantify and document real-world agreement rates.

Methodology

Full transparency on how our models work — from CV parsing to interview question generation and candidate scoring.

Industry

Applied research on hiring trends, time-to-hire patterns, and AI adoption across industries and company sizes.

What the data shows.

Two of our most-cited findings — independently validated and regularly updated with each model release.

Accuracy · Feb 2026

Human–AI Score Agreement

Recruit v2 scores compared against 42 senior HR professionals across 500 candidates in 12 industries. Pearson correlation: r = 0.91.

91%agreement
Technical roles93%
Behavioural89%
Cultural fit86%

Bias & Fairness · Mar 2026

Bias Reduction After Audit

Following our Q1 2026 bias audit, all four measured dimensions improved significantly. Orange = post-audit; faded = pre-audit baseline.

Gender97% equitable
Age94% equitable
Name origin96% equitable
Language92% equitable
Before auditAfter audit
Quarterly bias audits are now standard. Next audit scheduled Q3 2026.

Publications

DATECATEGORYTITLE
Mar 2026Bias & Fairness
Gender Neutrality in AI Interview Scoring
Feb 2026Accuracy
Human HR vs AI Score Agreement: Recruit v2 Benchmark
Jan 2026Methodology
How Recruit Scores a CV: A Full Walkthrough
Apr 2026Industry
2025 Hiring Trends: Data from 500 Companies
Mar 2026Methodology
Follow-up Question Generation: Design Principles
Feb 2026Bias & Fairness
Eliminating Name and Age Signals from CV Scoring
Jan 2026Accuracy
Interview Completion Rate: What Drives Candidates
Dec 2025Methodology
Structured vs Unstructured Interviews: Why We Chose Structure

Questions about our
methodology?

We're open to academic collaboration, external audits, and dataset sharing requests.

Open to collaboration

Dataset access for academic researchers
External bias audits welcome
Methodology review & peer feedback
Joint publication consideration
IRB-compatible data sharing agreements
All data requests reviewed within 5 business days.