Skip to content

Fairness and human review

Structured evidence can support better review. It cannot guarantee fairness.

ProofOfFit is designed to help people organize job criteria and candidate evidence in a more reviewable form. It does not claim to eliminate bias, prevent discrimination, certify legal compliance, or replace accountable human judgment.

Last reviewed: August 7, 2026

Design principles

Start with stated job criteria

ProofOfFit is designed to organize candidate information around criteria defined for a role. The quality and lawfulness of those criteria still require human review.

Show the evidence being considered

Reviewers should be able to inspect the resume, project, work-history, or other evidence used to support an assessment instead of relying on a score alone.

Keep people responsible

AI-assisted outputs are drafts and decision support. Employers and candidates remain responsible for checking accuracy, context, accommodations, and final decisions.

Important limitations

  • Software cannot guarantee that a hiring process is unbiased, fair, or legally compliant.
  • Job criteria can themselves encode unnecessary or discriminatory requirements and should be reviewed before use.
  • Protected characteristics and correlated information can appear directly or indirectly in resumes, work history, geography, education, or other evidence.
  • AI-assisted extraction, classification, and drafting can be incomplete or wrong and must be reviewable and correctable.
  • ProofOfFit does not currently publish a validated percentage reduction in bias, legal risk, hiring time, compliance cost, or demographic disparity.

Responsible use

Fairness depends on the full hiring process, including role design, sourcing, accommodations, interviews, decision rules, human behavior, and local law. Software is only one part of that system.

  1. 1Use job-related criteria that are necessary and appropriate for the role.
  2. 2Review criteria and outcomes for unintended exclusion or proxy effects.
  3. 3Provide appropriate human review, accommodations, correction paths, and appeals where required.
  4. 4Do not use a ProofOfFit score or generated draft as the sole basis for an employment decision.
  5. 5Apply the employer's legal, privacy, recordkeeping, and equal-opportunity obligations independently of the software.

Evidence and claim status

ProofOfFit does not publish a fairness-effect percentage on this page because a defensible claim would require a defined evaluation method, population, protected-class handling, baseline, sample size, uncertainty analysis, limitations, and current review.

We also do not present EEOC, GDPR, CCPA, SOC 2, ISO 27001, or similar labels here as proof that a hiring outcome is fair or that the product is legally compliant. Security certifications, privacy controls, and legal obligations have different scopes and require separate evidence.

When validated fairness evaluations are available for publication, their method, scope, results, and limitations should be published together rather than reduced to an unsupported marketing statistic.

Related information

This page describes product design intent and limitations. It is not legal advice and does not guarantee any employment, fairness, compliance, or regulatory outcome.