THE ROSAI FRAMEWORK

AI capability is not absolute. It depends on the role.

The AI skills expected from a Product Manager are different from those expected from an Engineer, Data professional, Transformation leader, Designer, or Business Head.

And increasingly, those roles overlap.

A Product Manager may be expected to prototype. An Engineer may need product judgement. A Transformation leader may need enough technical depth to evaluate AI solutions. Different organizations also expect very different combinations of these capabilities from people carrying the same job title.

ROSAI is designed for this reality.

ROSAI stands for Role-Oriented Skills Assessment for AI.

The framework evaluates AI readiness relative to the mix of responsibilities expected from the role. As the role orientation changes, ROSAI dynamically adjusts the importance and weighting of individual AI capabilities.

The result is therefore not simply a measure of “How much AI does this person know?”

It is a measure of:

How well does this person’s AI capability fit the AI expectations of this role?

Why SCORE-AI exists

Almost everyone can now claim to be AI-ready, AI-skilled, or even an AI expert.

The harder question for a hiring manager is:

Expert at what — and is that the expertise this role actually needs?

Someone may be highly capable at building AI workflows but weak at evaluating risk. Another person may have excellent AI product judgement without being a deep technical builder. Both may legitimately be strong AI practitioners, but their suitability changes depending on the role being assessed.

SCORE-AI uses the ROSAI Framework to convert these different capability signals into a role-oriented readiness profile.

Role-oriented, not role-agnostic

The candidate defines the expected mix across areas such as:

Product & Business · Engineering · Data & ML · Transformation · Design & UX · Leadership & Governance

ROSAI then adjusts how strongly different AI capabilities contribute to the final score.

The same person can therefore have a different readiness profile for two different roles — because the expectations themselves are different.

This makes the ROSAI Score more like a dynamic unit of measurement than a fixed test score.

The reference point changes with the role.

Evidence without placing the burden on the interviewer

Claims alone are not enough.

Candidates may support their experience with evidence such as:

• live products or prototypes
• repositories
• demos
• architecture documentation
• case studies
• measurable business outcomes
• other relevant work

SCORE-AI evaluates these signals using a structured evidence framework and reports a separate Evidence Strength measure.

This gives HR and leadership teams an additional, standardized signal without requiring them to independently determine how much weight to place on every portfolio link, technical artifact, or project claim.

What the ROSAI Score tells you

The ROSAI Score is intended to answer a more useful hiring question than:

“Is this person an AI expert?”

Instead:

“How well does this person’s demonstrated AI capability align with what this specific role requires?”

That distinction is the foundation of the ROSAI Framework.

Seven dimensions. Your role as the context.

Role-adjusted scoring

You allocate your role across six areas. That mix determines the importance of each capability. Seniority does not directly add points. Serious gaps in a core capability can limit the overall score, so a weighted average does not conceal an important weakness.

Three outputs, read together

• ROSAI Score: a 0–100 role-adjusted capability signal, combining structured answers and rubric-graded scenarios.
• Evidence Strength: availability of distinct, self-declared evidence sources. It does not mean those claims have been independently verified.
• Assessment Confidence: an interpretive indicator based on completion, consistency, evidence, scenario alignment, and response quality.

Evidence Strength and Assessment Confidence do not modify the ROSAI Score.

How the assessment works

Answer 18 questions, one screen at a time. Begin with role mix and responsibility level, then explore usage, judgement, experience, evidence, and impact. Finish with three short scenarios selected from your role mix. Each scenario accepts 5–700 characters; concise, specific reasoning is welcome.

The AI evaluator grades only the three scenario answers against a defined rubric. Structured scoring remains deterministic. Grader outputs and framework versions are retained with each result; existing results are not silently rescored under new versions.

Understanding the bands

What this experiment does not claim

SCORE-AI is not a certification, a hiring recommendation, or a psychometrically validated instrument. It does not provide population percentiles, detect AI-generated writing, independently verify evidence, or predict job performance. The current experiment is strictly non-commercial and for personal use.

Privacy by default

Your answers are private. You can enable and revoke a share link and separately choose whether to include evidence details. Research, grader-improvement, and future-training permissions are optional. Read the privacy notice.

Frequently asked questions

Framework ROSAI 1.0 · Evidence availability revision 1.1