RSIT's Patent-Pending Capstone Architecture

Not a quiz. Not a module.
A governed intelligence assessment.

Every RSIT certification is awarded through a fully patent-pending autonomous capstone system. Students construct 8 governed intelligence artifacts, pass a binary structural gate, score across 7 weighted dimensions, and attest to four constitutional doctrines. This page explains exactly how it works.

8
Governed Artifacts
12
Structural Gate Checks
7
Scoring Dimensions
70%
Pass Threshold
4
Constitutional Doctrines
0
Human Reviewers
How it flows

The Assessment Pipeline

Every submission passes through the same deterministic sequence. No exceptions. No human override.

1
Student Builds
8 Artifacts
AI-guided
2
Structural
Gate Runs
12 binary checks
×
Gate Fail
Blocked. No score.
Gate Pass
Scoring unlocked
3
7-Dimension
Scoring
Weighted & deterministic
4
Pass ≥70%
or Retry
5 attempts max
Certification
Awarded
RSCFAI-attested
The Deliverables

The 8 Governed Intelligence Artifacts

Every certification, every domain, every student produces exactly these 8 artifacts. Structure is universal. Content is domain-specific per the governing MIP.

ARTIFACT 01
Synthetic Scenario Definition
Establishes the fictional organizational context, AI system type, jurisdiction, and operating environment for the entire capstone. All downstream artifacts reference this scenario.
Fully Fictional — Synthetic-Only Doctrine
ARTIFACT 02
Risk & Pattern Matrix
Maps AI-specific risk patterns to derived risk levels using locked enumerations. Requires minimum 20 rows. Each row derives risk deterministically from canonical pattern classifications — not from judgment.
Minimum 20 rows · Locked enumerations
ARTIFACT 03
Governance & Decision Boundary Schema
Defines the decision logic governing every AI action in the scenario: what the system is permitted to decide, what requires human review, what is escalated, and what is blocked — mapped to AccountableRole and DecisionStatus enumerations.
4 locked DecisionStatus values
ARTIFACT 04
Compliance Mapping & Regulatory Alignment Table
Cross-maps the scenario's governance requirements to applicable regulatory frameworks, standards, and jurisdiction-specific obligations. Domain-specific per the governing MIP.
Domain-specific via MIP regulatory overlay
ARTIFACT 05
Trust, Ethics & Guardrail Definition
Deploys all 15 canonical guardrails (G-01 through G-15) against the scenario. Each guardrail must be present, mapped to a specific risk, assigned activation conditions, and linked to an accountable role.
All 15 guardrails G-01–G-15 required
ARTIFACT 06
Metrics & Monitoring Framework
Defines all 12 canonical metrics (M-01 through M-12) with scenario-specific thresholds, monitoring cadences, and escalation triggers. Each metric must be calibrated to the scenario context.
All 12 metrics M-01–M-12 required
ARTIFACT 07
Orchestration Blueprints / Layer Definitions
Maps the AI system's operation across the 6 canonical orchestration layers: Sense → Constrain → Decide → Observe → Attest (L1–L6). Each layer must be populated with scenario-specific governance controls.
6 layers L1–L6 · Full coverage required
ARTIFACT 08
Contribution & Self-Attestation Record
The student's constitutional self-attestation: confirming adherence to all four doctrines, identifying the governing MIP version, and attesting that the submission is original. Creates the DEAL™ attestation chain.
DEAL™ timestamped · MIP version logged
The First Filter

The Structural Gate

Before any scoring occurs, every submission passes through a binary hard-fail structural gate. The gate runs 12 checks. Each check produces a pass or fail. A single failure blocks the submission entirely — no partial credit, no override, no exceptions.

This is not a rubric. It is a technical enforcement mechanism. The gate runs deterministically, in the same order, every time. No human is involved in the gate evaluation.

Only submissions that pass all 12 checks proceed to scoring.

Why this matters

The gate enforces the single most important principle in AI governance: completeness is non-negotiable. An AI system with an incomplete guardrail catalog is not a well-governed AI system — it is an ungoverned one.

01
Artifact Count — Exactly 8 artifacts present.
02
Template Integrity — No columns renamed, merged, added, or removed.
03
Row Density (A2) — Risk & Pattern Matrix contains minimum 20 substantive rows.
04
Enumeration Compliance — All locked enumerations use only canonical values.
05
Risk Derivation Logic — DerivedRiskLevel follows the deterministic derivation rule.
06
Decision Derivation — DecisionStatus follows the deterministic logic rule.
07
Synthetic-Only Doctrine — No real companies, people, events, or data.
08
No Code Doctrine — No code, pseudocode, SQL, API calls, or system diagrams.
09
No Advisory Doctrine — No “should implement,” “best practice recommends,” or advisory language.
10
Guardrail Coverage — All 15 canonical guardrails G-01 through G-15 present.
11
Metric Coverage — All 12 canonical metrics M-01 through M-12 present.
12
Cross-Artifact Traceability — Risk IDs referenced consistently across A3, A5, A6, A7.
After the gate passes

The 7 Scoring Dimensions

Scoring runs only after all 12 structural gate checks pass. Weights are identical across every certification and every domain.

Pattern Coverage
Breadth and accuracy of canonical governance pattern application
15%
Risk Derivation
Correctness of DerivedRiskLevel assignment against the deterministic rule
15%
Decision Coherence
Logical consistency of DecisionStatus assignments
15%
Guardrail Strength
Quality and specificity of guardrail configuration
15%
Metric Integrity
Accuracy of metric thresholds, cadences, and escalation triggers
15%
Traceability
Cross-artifact consistency of references across all 8 artifacts
15%
Evidence Coverage
Completeness of evidence object references supporting governance claims
10%
70%
Passing Threshold

Required across the weighted composite of all 7 dimensions. No domain-specific scoring adjustment.

Why 70%? The threshold reflects the minimum demonstrated competency a practitioner needs to apply governance frameworks in a real deployment context.
6 dimensions @ 15%90%
1 dimension @ 10%10%
Total possible100%
When a submission does not pass

The Retry Policy

Every student receives structured, escalating support. The system distinguishes between diagnostic feedback and guided correction — and enforces a hard limit.

1
Diagnostic Mode
Full scoring report with dimension-level breakdown and specific feedback.
2
Diagnostic Mode
Pattern comparison against prior submission. Identifies addressed vs. new issues.
3
Diagnostic Mode
Trend analysis across all three attempts. Persistent confusion patterns flagged.
4
Guided Correction
Row-by-row guided correction. System identifies specific derivation errors.
5
Guided Correction
Final attempt. Full row-by-row correction with canonical reference citations.
🔒

After 5 Failed Attempts: Mandatory Remediation

Hard lockout. All 4 remediation modules required at 90% minimum each. Upon clearance, a new scenario is generated — the attempt counter resets to zero.

After lockout

The 4 Remediation Modules

Each module targets a specific failure domain. All 4 must be completed at 90% minimum — higher than the capstone itself.

01
Risk Derivation
Addresses the most common structural failure: applying judgment instead of following the deterministic derivation rule.
→ Minimum 90% to unlock
  • DerivedRiskLevel derivation logic
  • Boolean flag interpretation
  • Pattern classification mapping
  • Edge case disambiguation
02
Decision Logic
Rebuilds DecisionStatus derivation — conditional logic mapping risk level and flags to Approved, Human-Reviewed, Escalated, or Blocked.
→ Minimum 90% to unlock
  • DecisionStatus enumeration logic
  • Conditional derivation rules
  • Cross-row coherence
  • Escalation vs. block thresholds
03
Guardrail Mapping
Addresses incomplete guardrail deployment — the second most common gate failure. All 15 canonical guardrails covered.
→ Minimum 90% to unlock
  • G-01 through G-15 functions
  • Activation condition logic
  • AccountableRole assignment
  • Domain-specific guardrail priority
04
Traceability
Builds cross-artifact consistency. Risk IDs, guardrail references, and metric links must resolve across A2, A3, A5, A6, and A7.
→ Minimum 90% to unlock
  • Risk ID referencing across artifacts
  • Guardrail cross-reference consistency
  • Metric linkage to risk rows
  • Orphaned reference detection
Constitutional constraints

The Four Doctrines

Every capstone submission is bound by four constitutional doctrines. Violation of any doctrine fails the structural gate — regardless of content quality.

Doctrine I
Synthetic-Only
All capstone work must be fully fictional. No real companies. No real people. No real data. No real events.
The synthetic requirement ensures the assessment evaluates governance design — not research ability.
Doctrine II
No Code
No code, pseudocode, scripts, SQL, API calls, or system diagrams implying implementation.
RSIT certifications assess governance competency, not engineering competency.
Doctrine III
No Advisory
No advisory language. Prohibited: “should implement,” “best practice recommends,” “we advise.”
Governance frameworks are operational constraints, not recommendations.
Doctrine IV
Template Fidelity
Students fill content only. No columns added, renamed, merged, or removed. No artifacts combined or split.
Schema immutability is what makes cross-cohort comparison and the intelligence flywheel possible.
The Governing Document

The RSCFAI Master Implementation Plan

Every certification is governed by a Master Implementation Plan (MIP) authored and version-controlled by RSCFAI. The MIP parameterizes the entire capstone: which regulatory frameworks apply, which patterns receive emphasis, which guardrails are prioritized, and which scenario types are valid.

The architecture is layered: one Universal Core MIP applies to all 117+ certifications. Six Domain MIPs add the regulatory overlay specific to each industry domain.

RSIT cannot modify a MIP. Only RSCFAI holds that authority. This separation is constitutional — the entity that delivers the certification cannot change the standards against which it delivers.
Domain MIPIndustry CoverageCertifications
Universal CoreAll certifications — base schemas, enumerations, scoring, gate rules, doctrines117+
D1 — AI EngineeringAI development, robotics, autonomous systems, ML Ops, agent architecture~17
D2 — Financial & LegalBanking, investment, insurance, regulatory compliance, legal tech~10
D3 — Healthcare & Life SciencesClinical AI, pharma, longevity, medical devices — dual clinical/non-clinical routing~26
D4 — Business OperationsSupply chain, operations, HR tech, manufacturing, enterprise AI~32
D5 — Sports & MediaSports analytics, content AI, media platforms, audience intelligence~10
D6 — Aesthetics & PersonalizationBeauty tech, fashion AI, personalization engines, recommendations~21
Upon Certification

5 Career-Ready Deliverables

Every certified professional receives a complete, cryptographically attested credential package — automatically generated the moment you pass.

Document 02

Certificate of Completion

Formal credential with double-border design, overall score, structural gate status, and embedded QR code linking to public verification.

PPTX & PDF · Landscape · QR Embedded
Document 03

Employer Summary Deck

4-slide presentation with scoring bars, competency proofs, regulatory mapping, and QR verification. Designed for hiring managers and compliance teams.

PPTX & PDF · 4 Slides · Executive
Document 04

LinkedIn Badge

1080×1080 image optimized for LinkedIn feed posts. Name, certification, score, and QR code — download, post, and let employers scan to verify.

PNG · 1080×1080 · LinkedIn Optimized
Document 05

Presentation Badge

Square-format slide for conference decks, digital portfolios, and presentations. Fully editable in PowerPoint or Google Slides.

PPTX · Square 7.5″×7.5″ · Editable
Every document includes a unique SHA-256 attestation hash and scannable QR code — cryptographically verifiable by any employer, partner, or institution.
For Employers & Institutions

Verify Any Credential in Seconds

Every RSIT credential is cryptographically attested with a unique SHA-256 hash and independently verifiable through our public portal. No login. No phone calls. No waiting.

Whether you’re a hiring manager, compliance officer, or university partner — verification takes seconds.

  1. Find the completion code on any certificate, portfolio, badge, or deck
  2. Enter the code on our verification page — or scan the QR code
  3. Instantly see name, certification, full scoring breakdown, and SHA-256 hash
Verify a Credential →
shermanpathways.com/pages/verify
Credential Verified
Ivan Del Valle
Certified Strategic AI Agent Practitioner (CSAAP)
94%
DomainD1: AI Engineering
GatePASS
DateMarch 17, 2026
Assessment7-Category Weighted
SHA-256 Attestation
9e6bd2b8a430816cb832590740fc58f879afe954121991cfc9609f755474ff14
Ready to build

This is how practitioners are made.

Not through modules and a final quiz. Through the construction, validation, and constitutional attestation of 8 governed intelligence artifacts — assessed against the same deterministic standard that governs every cohort before you.