Q3 Institutional

Q3 Institutional (ROAI 4-Category Framework v2026.1)

Position in the Framework

Q3 Institutional is the third category in the ROAI 4-Category Framework. It measures the operating-model muscle around AI: the governance, people, cadences, and telemetry that persist even when specific tools or models are replaced.

What Q3 Measures

Q3 captures institutionalized capability that is:

• Tool-agnostic – survives vendor and model changes.
• Embedded in the operating model – shows up in charters, roles, and recurring meetings.
• Cumulative over time – each cycle of the methodology adds more Q3 value.

Core components:

1. Named Owners and Decision Rights
   • AI Council members with explicit scopes.
   • Risk Register owners for each material risk domain.
   • Use Case owners accountable for lifecycle and controls.
   • Vendor relationship owners with clear authority and escalation paths.
2. Cadence Integrity
   • Quarterly Cadence Retrospective (GOV-15) runs on schedule and produces actions.
   • Monthly AI Council meets with quorum, agenda, and minutes.
   • STR-07 Annual Charter Refresh executed each year, with deltas vs. prior year.
   • Other recurring forums (e.g., model review boards, risk sign-off meetings) that actually occur, not just exist on paper.
3. Trained Staff
   • Practitioners and owners who have completed canonical training on:
     • The AI governance methodology.
     • Specific capabilities (e.g., GenAI assistant, decisioning engine, retrieval system).
   • Documented currency (e.g., re-certification, refreshers, versioned training content).
4. Telemetry Infrastructure
   • Tool-agnostic measurement plumbing:
     • Drift and stability telemetry.
     • Output-quality sampling and review workflows.
     • Hallucination-rate or error-rate monitoring where applicable.
     • ROAI measurement pipelines and dashboards.
   • Designed so that when tools change, telemetry endpoints are re-wired, not re-invented.
5. Vendor and Partner Relationships
   • Named contacts at vendors, integrators, and assurance partners.
   • Clear engagement models (RACI, SLOs, escalation paths).
   • Ability to re-activate relationships quickly when new capability is needed.

Why Q3 Compounds Slowly and Persists

• Cadence-driven accretion – Each completed cycle (monthly council, quarterly retrospective, annual refresh) adds:
  • More refined roles and decision rights.
  • Better-documented processes and playbooks.
  • Improved telemetry and reporting.
• Time as a multiplier – Two years of running the methodology ≈ twice the Q3 value of one year, even if Q1 (direct ROI) looks similar.
• Persistence across change:
  • Vendor or model swaps do not reset governance; owners, cadences, and methods remain.
  • Historical decisions and risks remain accessible via the Risk Register and AI Council Decision Log (GOV-13).
  • New capabilities can be onboarded faster because the function reuses existing owners, cadences, and telemetry patterns.

Where Q3 Lives in the Operating Model

• AI Governance Charter (GOV-01)
  • Explicitly names Q3 capability accretion as an objective (e.g., “Increase institutional AI operating capability year-on-year”).
• STR-07 Annual Charter Refresh
  • Primary instrument for year-over-year Q3 measurement.
  • Compares:
    • Number and clarity of named owners.
    • Cadence adherence and scope.
    • Trained staff counts and coverage.
    • Telemetry maturity and coverage.
    • Vendor/partner landscape and depth.
• Quarterly Cadence Retrospective (GOV-15)
  • Maintains the running Q3 inventory:
    • Which cadences ran or slipped.
    • Which roles were added, changed, or left vacant.
    • Which telemetry gaps were closed or opened.
• AI Council Decision Log (GOV-13)
  • Records creation, modification, and sunset of Q3 elements:
    • New recurring meetings or forums.
    • New owner assignments or reassignments.
    • New training cohorts and certification standards.
    • Decommissioned capabilities and how their records are archived.

Common Failure Modes

1. Q3 Attributed to Tools
   • Misconception: “The tool gives us governance.”
   • Reality: Governance is provided by people + process + telemetry, not by the vendor product.
   • Symptom: When the tool is replaced, governance collapses.
2. Q3 Dependent on Individuals
   • Single experts hold critical knowledge with no documentation.
   • Departure or role change resets the Q3 baseline.
   • Mitigation: Playbooks, runbooks, RACI matrices, and cross-training.
3. Q3 Ungoverned
   • No maintained inventory of:
     • Cadences and their health.
     • Named owners and coverage.
     • Trained staff and currency.
   • Gaps are discovered only under audit, incident, or regulator scrutiny.

Relationship to DE-4 and DE-5

A function that maintains a running Q3 inventory and revisits it during STR-07 Annual Charter Refresh simultaneously evidences:

• DE-4 (Governance Posture) – Clear roles, cadences, and decision records.
• DE-5 (Continuous Learning) – Systematic improvement of the operating model over time.

In practice, Q3 is the category that makes AI capability defensible, auditable, and repeatable, independent of any specific vendor or model.