AI Transparency
How Carevo's AI makes decisions, with details on fairness and bias mitigation.
AI Transparency
Carevo uses AI extensively to power the Career Agent, match scoring, and tools. We believe you should understand how these systems make decisions that affect your job search or hiring process.
How AI Decisions Are Made
Match scoring — When a job is scored against your Career DNA, the AI evaluates six dimensions independently. Each dimension uses a combination of vector similarity (for fast initial matching) and detailed prompt-based analysis (for deep fit scoring). The AI reads the full job description and your complete Career DNA to generate scores and explanations.
Agent conversations — Your Career Agent uses Claude by Anthropic to generate responses. It receives your Career DNA, recent conversation history, and relevant context (like a job listing you are discussing) as input. Responses are generated in real time and are not pre-scripted.
Tool outputs — Tools like the Resume Analyzer and JD Analyzer use AI to evaluate documents. The AI follows structured evaluation rubrics to ensure consistent and fair assessments.
Bias Mitigation
We take several steps to reduce bias in AI decisions:
- Structured scoring rubrics — The AI evaluates against defined criteria rather than making open-ended judgments. This reduces the influence of implicit biases in language models.
- Blind matching — During the signal and deep fit stages, candidate and employer identities are not visible. Matching is based purely on skills, preferences, and fit dimensions.
- Regular audits — We periodically analyze match and scoring outcomes across demographic groups to identify and correct any systematic biases.
- No demographic scoring — The AI is explicitly instructed not to factor in age, gender, race, ethnicity, religion, disability status, or other protected characteristics.
Explainability
Every match score comes with an explanation. You can ask your Career Agent to break down why a particular score was given, which factors contributed most, and what would need to change for the score to improve. Employer screening reports also include reasoning for each evaluation.
Human Oversight
AI decisions on Carevo are advisory, not final. You always have the ability to override any recommendation, edit your Career DNA to adjust future scoring, or contact support if you believe a score is incorrect. The "Review before submit" option on auto-apply ensures a human is always in the loop for application decisions.
Model and Data
Carevo uses Anthropic's Claude models. Your data is sent to Anthropic's API for processing but is not used to train their models. See our Privacy Policy for complete details on data handling with third-party AI providers.