WHERE THIS SITS IN THE ENGAGEMENT MODEL.
Stage 3 follows the Assessment (Stage 1) and Scenario Planning (Stage 2). Not every institution reaches Stage 3 — and not every institution should.
Stage 3 is engaged when:
- The Assessment has produced a 5×5 Control Matrix score and identified the specific governance gaps requiring architectural remediation
- The Strategic Assessment has produced a 0-160 score in the Compose (81-120) or Build (121-160) band
- Scenario Planning has stress-tested the strategic direction against multiple plausible futures and confirmed the architectural commitment is resilient
- The institution has secured board-level commitment, capital allocation, and operational readiness for a multi-year programme
Most institutions complete only Stages 1 and 2. The Assessment and the strategy program built around it are sufficient for institutions whose AI Control posture is strong or whose strategic direction does not require institution-owned infrastructure.
Stage 3 is the deep commitment. It is reserved for institutions where AI is mission-critical, regulatory or fiduciary obligations require demonstrable infrastructure ownership, and the institution is prepared to build, own, and operate AI control architecture as long-term institutional capability.
WHAT AI IMPLEMENTATION BUILDS.
THE INSTITUTIONAL AI STACK™ — DEPLOYED.
The five AI ecosystems — Power, Compute, Data Centers, Models, and Agents — designed, integrated, and deployed under one institutional control structure calibrated to the institution's specific regulatory, operational, and capital context.
What gets built: Custom architectural specification → vendor and infrastructure selection → deployment across the five ecosystems → integration testing → operational handover.
OLTAIX™ — CONFIGURED.
The control plane — orchestration, evidence, agentic governance, real-time compliance, and board-ready intelligence — configured to the institution's policy framework, governance structure, and reporting requirements.
What gets built: Policy translation into technical configuration → access control architecture → audit trail specification → board reporting structure → integration with the deployed Stack.
AI CONTROL — THE OUTCOME
Not just architecture — the institutional capability to govern, operate, and evolve AI control over time. Board reporting structure, audit protocols, escalation pathways, and ongoing governance reviews structured for fiduciary oversight and regulatory examination.
What gets built: Governance committee structures → board reporting cadence → audit trail accessibility for regulators → escalation protocols → ongoing programme evolution methodology.
THE IMPLEMENTATION SEQUENCE.
Four phases. 18–36 months. Defined milestones, board-quality deliverables, governance reviews at every transition.
PHASE 1 — DESIGN.
Duration: 3–6 months
What happens: Translation of strategic decision into architectural specification. The Stack design is calibrated to the institution's regulatory framework (ERISA, DORA, SEC, Solvency II, NAIC, or applicable). OLTAIX™ control plane configuration is mapped to institutional governance, policy, and reporting requirements. Vendor selection advisory begins.
Institutional involvement: Bi-weekly working sessions with technology, risk, and governance leadership. Monthly board or steering committee reviews.
Deliverables at end of phase: Complete architectural specification document, vendor selection recommendations, OLTAIX™ configuration map, multi-year deployment plan, and capital deployment schedule.
PHASE 2 — DEPLOY.
Duration: 6–12 months
What happens: Infrastructure deployment across the five ecosystems. Vendor contracts executed under institutional control. OLTAIX™ configured and integrated with the deployed Stack. Initial workloads migrated under controlled conditions. Continuous testing and validation.
Institutional involvement: Weekly programme reviews during active deployment. Monthly executive steering committee. Quarterly board governance reviews.
Deliverables at end of phase: Operational Stack across all five ecosystems, OLTAIX™ control plane operational, initial workloads under institutional control, integration test results, deployment certification.
PHASE 3 — VALIDATE.
Duration: 3–6 months
What happens: Validation of the deployed environment against the institution's regulatory obligations and the original Strategic Assessment requirements. Governance protocols tested. Board reporting structure validated. Audit trail demonstration. First regulatory readiness review.
Institutional involvement: Monthly validation reviews. Coordination with internal audit, compliance, and legal. Board governance committee review of validation results.
Deliverables at end of phase: Validation report against original Assessment findings, regulatory readiness certification, governance protocol documentation, board governance briefing pack, transition plan to ongoing operations.
PHASE 4 — OVERSEE.
Duration: Ongoing — typically 24+ months following deployment
What happens: Quarterly governance reviews ensure standards hold as AI deployment grows. Architecture and OLTAIX™ configuration evolve with regulatory developments and institutional priorities. New AI use cases evaluated against the established control framework. Annual re-benchmarking against the 5×5 Control Matrix.
Institutional involvement: Quarterly programme governance reviews with institutional leadership and Institutional AI senior team. Annual board-level review of AI control posture.
Deliverables at each cycle: Quarterly governance report, annual re-benchmarked Assessment, annual board-quality AI control posture report, ongoing strategic guidance on emerging AI control questions.
WHO'S INVOLVED.
Implementation engagements bring together institutional leadership, the Institutional AI senior team, and selected specialists.
FROM THE INSTITUTION.
Stage 3 implementation requires sustained executive sponsorship and operational involvement across multiple functions:
- Executive sponsor — typically the CEO, COO, or CIO carrying overall accountability for AI control
- Programme owner — senior executive responsible for day-to-day institutional coordination
- Technology and security leadership — CIO, CISO, infrastructure architects engaged throughout deployment
- Risk and compliance leadership — Chief Risk Officer, Chief Compliance Officer, legal counsel involved at validation gates
- Board engagement — quarterly governance reviews with audit, risk, or technology committee
WHAT STAYS WITH THE INSTITUTION.
Implementation produces institutional capability that does not depend on Institutional AI's continued involvement.
THE ARCHITECTURE. The Institutional AI Stack™ is designed for the institution and owned by the institution. No subscription dependencies. No license terminations. No vendor lock-in to Institutional AI itself.
THE CONTROL PLANE. OLTAIX™ is configured for the institution and operates under the institution's control. Configurations, policies, and audit trails belong to the institution.
THE GOVERNANCE PROGRAMME. Governance protocols, board reporting structures, audit trails, and escalation pathways are documented and transferable. Institutional staff are trained and capable of operating the programme independently.
THE RELATIONSHIP. Phase 4 ongoing oversight is a continuing engagement that institutions extend at their option — not a dependency the institution cannot exit. The architecture, the control plane, and the governance programme function regardless of whether Institutional AI continues to provide quarterly oversight.
The institution that completes Stage 3 owns its AI control posture permanently. Institutional AI's continued involvement is by mutual agreement — not by structural dependency.
ENGAGEMENT LOGISTICS.
How Stage 3 engagements are structured commercially and operationally.
COMMERCIAL STRUCTURE. Multi-year retainer engagement with milestone-based delivery. Fees are structured against the four phases described above with clear deliverables at each phase boundary. Phase 4 (ongoing oversight) operates on a different commercial structure — quarterly retainer at the institution's election.
CONFIDENTIALITY. Every aspect of the engagement is covered under non-disclosure agreement from Stage 1 forward. Architectural specifications, deployment details, and governance configurations are never referenced externally. Benchmark data is anonymized before incorporation into sector benchmarks.
GOVERNANCE. Quarterly governance reviews with the institution's audit, risk, or technology committee throughout the engagement. Annual board-level reviews following deployment. All reviews supported by structured, board-quality deliverables designed for fiduciary oversight and regulatory examination.
ENTRY POINT. No institution begins at Stage 3. Stage 3 is the architectural commitment that follows the Assessment (Stage 1) and, when relevant, Scenario Planning (Stage 2). Institutions interested in Stage 3 begin with the AI Control Assessment.
Specific engagement terms are documented in mutually executed engagement agreements. Specific fee structures, deliverables, and team configurations are calibrated to each institution's context.