RoleFox runs a structured AI pipeline that ingests your role catalogue, evaluates every role against a validated job architecture framework, and produces consistent, auditable profiles at enterprise scale.
The same methodology applied to every role in your organisation — not approximate, not session-dependent, not subject to prompt drift.
Every role profile passes through structured validation gates. Flags are named and specific. You know exactly why each role lands where it does.
Lever scores, archetypes, competencies, quality metrics — machine-readable output that feeds analytics, exports, and downstream decisions.
The pipeline runs locally on your infrastructure. AI generation uses the Google Gemini API — enterprise-grade, with no persistent data retention. Not a consumer product with undefined terms.
Each stage produces structured, validated output that feeds the next. No stage is skipped. No output is approximate. The pipeline runs deterministically — the same role produces the same profile.
It is the right question. Here is the direct answer.
| Dimension | RoleFox | General AI (ChatGPT, Claude) |
|---|---|---|
| Scale | ✓Processes your entire role catalogue in a single pipeline run. The same methodology applied to every role, regardless of volume. | ✗Context limits mean quality degrades after 20–30 roles. Each session starts fresh with no shared framework state. |
| Consistency | ✓The same role produces the same output. Deterministic stages, structured validation, auditable results. | ✗Output varies between sessions, between prompts, between models. No mechanism to guarantee identical treatment across roles. |
| Validation | ✓Each stage is validated against the framework. PASS / WARN / FAIL with named flags. You know exactly what each flag means. | ✗No validation layer. You read the prose and decide if it looks right. At scale, that is not a realistic option. |
| Structured output | ✓Lever scores, archetypes, competencies, quality metrics — structured data per role that feeds downstream analytics and exports. | ✗Produces text. Extracting and structuring that text is a second project. |
| Auditability | ✓Full audit trail in the output. You can show exactly why each role landed where it did — lever by lever. | ✗No audit trail. The conversation is ephemeral. The reasoning that produced a given output cannot be reproduced. |
| Framework fidelity | ✓The job evaluation framework is baked into the pipeline — not described in a prompt that drifts across sessions. | ✗The framework must be re-described in every session. Prompt length and model variance mean it is never consistently applied. |
| Data governance | ✓Pipeline runs locally on your infrastructure. AI generation uses the Google Gemini API — enterprise-grade, no persistent data retention, no training on your data. You control what runs and when. | ✗Data passes through a consumer API with broad, evolving terms of service. Most enterprise HR data cannot be shared under standard consumer terms without explicit legal review. |
The job evaluation framework at the core of RoleFox uses nine accountability dimensions to place every role on a defensible, comparable grade scale. Five levers discriminate across all thirteen grades. Three plateau at senior levels — their ceiling is real and accepted. Financial authority is calibrated to your organisation's specific financial context.
Upload your existing role data — job descriptions, internal codes, grade labels, department structure. RoleFox handles ingestion and grade mapping automatically.
Set lever weightings, financial calibration thresholds, and organisation-specific parameters. The universal framework is fixed. The calibration layer is yours.
Every role profile is available in the client portal with lever scores, quality indicators, validation flags, and export options. Review, adjust, and deliver.
Clients configure lever weightings, financial thresholds, and grade mapping within defined bounds. The methodology stays intact. The organisation gets genuine ownership of the outcome.
Apply a multiplier of 0.8× to 1.2× to any of the nine levers. A financial services organisation may upweight Cost of Failure. A technology company may upweight Complexity. The total score normalises back to the 65-point scale.
Populate revenue and budget ladders with your organisation's actual figures. Two tracks: revenue accountability for sales roles, expressed as a percentage of total revenue; budget accountability for all other functions.
Your internal grade labels map to the universal scale at ingestion. Client-facing output uses your terminology throughout. The universal framework operates underneath, invisibly.
The portal onboarding questionnaire walks through each configuration decision with plain-language explanations and a live score preview so clients see the effect of each choice before committing.
Lever scores, archetypes, competencies, quality metrics — machine-readable output per role that feeds analytics, exports, and rationalisation. Not just prose.
| Role | Grade | Track | Archetype | Quality | Status |
|---|---|---|---|---|---|
| Support Engineer I | G3 | IC | Technical | 95 | ● PASS |
| Support Engineer I | G3 | PL | Technical | 96 | ● PASS |
| Senior Support Engineer | G5 | IC | Technical | 96 | ● PASS |
| Principal Support Engineer | G7 | IC | Technical | 96 | ● PASS |
| Distinguished Support Engineer | G9 | IC | Technical | 86 | ● WARN |
RoleFox is not a tool you prompt. It is a pipeline you run. The difference shows in every output — consistent, validated, structured, and auditable across your entire role population.