AI Disclaimer

1. AI Technology Used

Clear Desk uses large language models (LLMs) to analyze resumes and provide screening insights. Here is how our AI technology works:

• AI Provider: We use Groq-hosted Llama models for all AI processing, including resume screening, job description generation, and interview question creation
• How It Works: When you upload a resume, our system extracts the text content, sends it along with your job description to the AI model, and receives a structured analysis including match scores, skill assessment, and recommendations
• Real-Time Processing: All AI analysis is performed in real-time. No resume data is stored by the AI provider after processing is complete
• Scoring: Candidates receive a match score from 0 to 100 based on skill alignment, experience relevance, and qualification fit against your job description

2. Limitations

While we strive for accuracy, our AI system has inherent limitations that users should be aware of:

• Text Extraction: AI may not perfectly extract or interpret all information from resumes, especially those with complex formatting, tables, or non-standard layouts
• Approximate Scoring:Match scores are approximate assessments, not absolute rankings. A score of 75 does not objectively mean a candidate is "75% qualified"
• Missed Information: AI may miss relevant experience, transferable skills, or domain expertise that is not explicitly stated in the resume
• Ambiguous Content: AI may misinterpret ambiguous resume formats, abbreviations, or industry-specific terminology
• Image-Based PDFs: Scanned or image-based PDFs cannot be processed. Only text-based PDF documents are supported
• Language: The AI performs best with English-language resumes. Accuracy may be reduced for resumes in other languages
• Context: AI lacks the contextual understanding that experienced recruiters bring, such as company culture fit, industry trends, or nuanced career trajectories

3. Bias Acknowledgment

We take AI bias seriously and actively work to minimize it, but no AI system is perfectly unbiased:

• Prompt Design: Our AI prompts are specifically designed to evaluate candidates solely on professional skills, experience, and qualifications — never on name, gender, age, race, religion, nationality, or any personal characteristic
• Inherent Limitations: Despite our best efforts, the underlying language models may contain biases from their training data. These biases can subtly influence scoring and analysis
• Human Review: We strongly recommend reviewing AI screening results for fairness before making hiring decisions, particularly for borderline candidates
• Reporting: If you observe any potential bias in AI results, please report it to us at bias@cleardesk.app. We investigate all reports and update our prompts accordingly

4. Recommendations

To get the best results from Clear Desk and ensure fair hiring practices, we recommend:

• Always review AI results: Use AI screening as a first-pass filter to surface top candidates, not as a final verdict
• Use AI as a filter, not a decision: Let AI narrow down a large applicant pool, but make final decisions based on human judgment
• Interview before deciding: Always conduct interviews with candidates before making hiring decisions
• Review borderline candidates: Pay special attention to candidates the AI may have underscored — they may possess valuable qualities not captured in their resume text
• Maintain legal compliance: Ensure your hiring process complies with all applicable employment laws, including anti-discrimination regulations
• Provide good job descriptions: The quality of AI screening is directly related to the quality and specificity of your job description

5. Accuracy

• We aim for 85%+ accuracy in matching candidates to job descriptions, but we do not guarantee this level of accuracy for every screening
• Accuracy depends on multiple factors including resume quality, job description specificity, and the relevance of the candidate pool
• We regularly review and improve our AI prompts and scoring methodology to increase accuracy
• User feedback plays a crucial role in our improvement process — we welcome your input on AI result quality
• For feedback or accuracy concerns, contact us at feedback@cleardesk.app