AI-Powered Role Architecture

Role profiles that are
built, not described

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.

rolefox — pipeline · nine stages · every role
Consistent across every role

The same methodology applied to every role in your organisation — not approximate, not session-dependent, not subject to prompt drift.

Validated, not assumed

Every role profile passes through structured validation gates. Flags are named and specific. You know exactly why each role lands where it does.

Structured data, not just prose

Lever scores, archetypes, competencies, quality metrics — machine-readable output that feeds analytics, exports, and downstream decisions.

🔒
You control what runs and when

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.


Nine stages.
Every role. No exceptions.

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.

Stage 00
Grade Calibration
Before any generation begins, the pipeline maps your grade labels to the universal 1–13 grade scale. Every subsequent stage operates against a consistent framework regardless of internal naming conventions.
client_grade: "Grade 5" universal_grade: 8 band: "Leadership" calibrated: true

Why not just use
ChatGPT?

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.

Nine levers.
Sixty-five points.
Thirteen grades.

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.

Discriminating · 0–10
Time Horizon
Score 1 → 10
Daily task cycle at the most junior grades through to generational organisational accountability at the most senior.
Discriminating · 0–10
Scope
Score 1 → 10
A single defined activity within one team through to full enterprise and ecosystem-level accountability.
Discriminating · 0–10
Decision Latitude
Score 1 → 10
Pre-authorised actions through to sole authority for the direction of the entire organisation.
Discriminating · 0–10
Influence
Score 0 → 10
Interaction limited to immediate colleagues through to the definitive external voice of the organisation.
Discriminating · 0–10
Impact
Score 1 → 10
Contribution within an immediate task through to the primary driver of enterprise value and trajectory.
Plateaus at Grade 8–9 · 0–5
Complexity
Score 1 → 5
A single established method through to frontier-level problems with no established precedent. Ceiling is real and accepted.
Plateaus at Grade 8–9 · 0–5
Cost of Failure
Score 1 → 5
Quickly correctable errors through to existential organisational risk. Once the ceiling is reached, there is no higher category.
Plateaus at Grade 8–9 · 0–5
Constraint Environment
Score 1 → 5
Fully prescribed procedures through to Board mandate and regulation only. Ceiling is real and accepted.
Qualitative
Financial Authority
Low → C-Suite
Qualitative band labels calibrated to the client's financial context via the companion Financial Calibration Tool.
Band Grades Score range Max / 65
Foundation 1 – 3
16
Practitioner 4 – 6
36
Leadership 7 – 9
53
Executive 10 – 11
61
C-Suite 12 – 13
65

From data to
deliverable in days.

Step 01

Ingest your role catalogue

Upload your existing role data — job descriptions, internal codes, grade labels, department structure. RoleFox handles ingestion and grade mapping automatically.

Accepts: XLSX, CSV, source JDs
Output: structured pipeline input
Grade mapping: automatic
Step 02

Configure the framework

Set lever weightings, financial calibration thresholds, and organisation-specific parameters. The universal framework is fixed. The calibration layer is yours.

Framework: universal 9-lever model
Weighting: bounded ±20% per lever
Financial: revenue and budget ladders
Step 03

Review in the portal

Every role profile is available in the client portal with lever scores, quality indicators, validation flags, and export options. Review, adjust, and deliver.

Exports: PDF, Word, Excel
Validation: PASS / WARN / FAIL
Portal: role explorer + dashboards

The framework is universal.
The calibration is yours.

Clients configure lever weightings, financial thresholds, and grade mapping within defined bounds. The methodology stays intact. The organisation gets genuine ownership of the outcome.

Lever Weighting

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.

£

Financial Calibration

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.

🎯

Grade Mapping

Your internal grade labels map to the universal scale at ingestion. Client-facing output uses your terminology throughout. The universal framework operates underneath, invisibly.

📋

Guided Configuration

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.

Framework configuration · active session
Time Horizon 1.0×
Scope 1.1×
Decision Latitude 1.0×
Influence 0.9×
Impact 1.2×
Complexity 1.0×
Adjusted total score previewing live Live preview

Structured data.
Every role. Every field.

Lever scores, archetypes, competencies, quality metrics — machine-readable output per role that feeds analytics, exports, and rationalisation. Not just prose.

sample pipeline output — role profiles Pipeline complete · all stages passed
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

Job architecture at a scale
no team could achieve manually

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.