AI Governance and Investment Roadmap — Asset Management Firm, Singapore
The Situation
A Singapore-based asset management firm managing SGD 4.8 billion in AUM had been watching competitors move into AI-assisted portfolio analysis and client reporting for 2 years. The firm's investment committee wanted to act. The compliance team wanted to know what MAS would say about it first.
That tension had kept the firm in planning mode for 14 months. Two internal working groups had been formed and dissolved. A vendor had been shortlisted and then quietly dropped when the compliance team raised questions nobody could answer about model explainability and audit trails.
The firm needed a strategy that satisfied both sides of the table. Commercial enough for the investment committee. Defensible enough for MAS.
What Amorisoft Did
Amorisoft structured the 12-week engagement in two parallel tracks that ran simultaneously from week 3 onward.
The first track was use case discovery and prioritisation. 26 stakeholder interviews across portfolio management, client services, risk, compliance, and operations produced 18 candidate AI use cases. Each was assessed on commercial value, data readiness, MAS regulatory sensitivity, and implementation complexity. 18 narrowed to 4 for the 2-year roadmap. The 2 year-one priorities were an AI-assisted client reporting generation tool and a portfolio risk anomaly detection model.
The second track was AI governance framework design. Amorisoft reviewed MAS guidelines on the use of AI and data analytics in financial services and mapped them to the firm's existing risk and compliance infrastructure. The governance framework covered model risk management, explainability requirements, human oversight protocols, audit trail standards, and vendor due diligence criteria. It was written to be submitted to MAS directly, not just used internally.
MAS reviewed the framework during week 11 of the engagement. Sign-off came in week 14, two weeks after the formal engagement closed. Amorisoft remained available for MAS queries during the review period at no additional charge.
The vendor selection framework delivered as part of the roadmap included evaluation criteria across 9 dimensions, a scoring template, and a shortlist of 3 vendors for each of the 2 year-one initiatives.
Results
MAS provided sign-off on the AI governance framework without requesting revisions. The investment committee approved the 2-year roadmap and year-one initiative funding at the same board meeting. Both year-one initiatives, client reporting generation and portfolio risk anomaly detection, went live within 8 months of the engagement closing. The governance framework has since been adopted as the firm's standing model risk policy, covering all AI and algorithmic tools across the business.