How AI Generates Your Scorecard

When you provide a job description to Waffle, the AI performs a multi-step analysis to create a comprehensive interview scorecard. Here’s what happens behind the scenes—and what you’ll receive.

​

Confirmation step

Before generating any scorecard, Waffle shows you a confirmation card summarizing what it’s about to build. The card includes:

• Role and interview type — the position title and format the AI detected
• Estimated competencies — how many competency areas the scorecard will cover
• Source description — whether the JD came from your paste or was auto-fetched from the web

You have two options:

• Generate Scorecard — proceed with generation using the current inputs
• Adjust first — go back and refine your requirements before generating

This step gives you a chance to verify the AI understood your intent before spending a generation credit. If anything looks off, click “Adjust first” and add more detail in the chat.

​

The AI analysis process

Waffle’s AI reads your job description and identifies the skills, attributes, and context that matter most for the role. It then generates:

1. Key competency areas based on role requirements
2. Importance weights (1-5 scale) reflecting how critical each competency is
3. Targeted interview questions with follow-ups for each competency
4. Behavioral scoring rubrics with anchor descriptions at each rating level
5. Role-specific red flags to watch for during interviews
6. EEOC compliance guidance to ensure legally sound interviewing

The entire scorecard typically streams in over 2-4 minutes (up to 6 minutes in ULTRA Mode), with results appearing in your chat live as the AI generates them — no waiting on a blank loading screen.

​

Real-time streaming visualization

While your scorecard is being built, Waffle displays a live chart that visualizes progress in real time. You can watch competencies, questions, and rubrics populate as the AI generates them — no loading spinners or waiting around. This streaming visualization is available on every plan, including Free. It appears automatically above your scorecard whenever you start a new generation.

​

Competencies

Every scorecard includes 4-8 competencies tailored to the specific role. Each competency represents a distinct skill or attribute the AI identified as important.

​

Importance Weights

Competencies are weighted on a 1-5 scale to signal priority:

• 1-2: Nice-to-have skills (lower priority)
• 3: Moderate importance (solid contributor trait)
• 4-5: Critical skills (make-or-break for success in the role)

For example, a “Senior Software Engineer” scorecard might have “Technical Problem Solving” weighted at 5/5, while “Project Management” is weighted at 3/5—both matter, but technical skills are more critical.

​

Typical Competency Examples

• Technical roles: Problem Solving, System Design, Code Quality, Collaboration
• Leadership roles: Strategic Thinking, Team Development, Stakeholder Management, Decision Making
• Design roles: User Research, Visual Design, Prototyping, Cross-Functional Collaboration

​

Interview Questions

For each competency, Waffle generates behavioral and situational questions designed to elicit evidence of the candidate’s abilities.

​

Question Structure

Each question includes:

• Primary question: The main prompt to ask the candidate
• Follow-up questions: Suggested probes to dig deeper into the candidate’s response

Example for “Technical Problem Solving” competency:

• Primary: “Describe a time you solved a complex technical problem. What was your approach?”
• Follow-ups: “What alternatives did you consider? How did you validate your solution?”

Questions are mapped to their corresponding competency, so interviewers know which skill they’re evaluating with each question.

​

Scoring Rubrics

Each competency includes a 1-5 behavioral rating scale with clear anchor descriptions at each level. These anchors help interviewers assign consistent scores based on what “good” and “great” actually look like.

​

Rubric Structure

• 1 (Needs Improvement): Minimal evidence of the skill; significant gaps
• 2 (Developing): Some evidence, but inconsistent or surface-level
• 3 (Meets Expectations): Solid, consistent demonstration of the skill
• 4 (Exceeds Expectations): Strong, nuanced demonstration with depth
• 5 (Outstanding): Exceptional mastery; clear evidence of excellence

Example for “Communication” competency:

• 3: “Communicates clearly in written and verbal formats; adjusts tone for audience”
• 5: “Articulates complex ideas simply; proactively seeks feedback; influences cross-functional stakeholders effectively”

These behavioral anchors reduce subjectivity and help multiple interviewers score candidates consistently.

​

Red Flags

Waffle identifies potential warning signs specific to the role. These aren’t automatic disqualifiers—they’re signals to probe further or consider carefully.

​

Typical Red Flag Examples

• For technical roles: Inability to explain past technical decisions; blaming others for project failures
• For leadership roles: Lack of self-awareness; poor listening skills during interview
• For design roles: No portfolio examples showing user research; defensive about design critique

Red flags give interviewers a checklist of behaviors to watch for, helping them spot potential mismatches early.

​

EEOC Compliance

Every scorecard includes legal guidance to help interviewers avoid discriminatory questions. Waffle provides:

• Questions to avoid: Topics like age, marital status, religion, disabilities (unless job-related)
• Job-relevant focus: Reminders to keep questions tied to role requirements
• Behavioral emphasis: Guidance to focus on past behavior and specific examples, not hypotheticals or protected characteristics

This built-in compliance layer helps hiring teams conduct fair, legally sound interviews that focus on job-relevant criteria.

​

AI competency scoring

When you attach an interview transcript to a scorecard, Waffle’s AI automatically scores the candidate against every competency. This turns your scorecard from a planning tool into a scoring tool — no manual evaluation required.

​

How it works

1. Upload or paste a transcript — attach a VTT, SRT, JSON, or plain text transcript to any scorecard (see uploading transcripts)
2. AI reads the full conversation — the model processes all speaker turns and evaluates them against each competency on the scorecard
3. Each competency receives a score — the AI assigns a rating using the same 1-5 rubric scale you see on the scorecard
4. Evidence excerpts are attached — up to three direct quotes from the transcript are linked to each competency score, so you can see exactly what the candidate said
5. Confidence levels are reported — each score includes a confidence indicator (low, medium, or high) based on how much relevant evidence was found

​

Confidence threshold

The AI only assigns a rating when it has sufficient evidence. If the confidence level falls below the threshold, the competency is left unscored rather than guessing. This prevents misleading scores when a transcript does not cover a particular skill area.

• High confidence — strong, direct evidence in the transcript
• Medium confidence — some relevant evidence, but not definitive
• Low confidence — minimal evidence found; no rating assigned

​

Scoring and EEOC compliance

The scoring engine follows EEOC guidelines throughout the evaluation. It is designed to:

• Evaluate only job-relevant competencies listed on the scorecard
• Ignore protected characteristics — race, gender, age, religion, disability, national origin, pregnancy status, and genetic information are excluded from scoring
• Base ratings on demonstrated behavior — scores reflect what the candidate said and did, not assumptions or inferences about identity

​

Token budget

Transcript scoring consumes tokens from your plan’s monthly budget. Each scoring run uses tokens proportional to the transcript length and number of competencies. You can check your current usage in Settings → Billing. If your budget is exceeded, scoring pauses until your next billing cycle. You can upgrade your plan for a higher budget.