From ad-hoc prompts to a governed AI research capability
Analysts were experimenting with AI without a firm-wide framework. PICOLens set the AI strategy, wrote the Responsible AI policy and model governance, and ran hands-on bootcamps. The board approved the policy and the investment committee funded three research agents.
- Client
- Singapore-headquartered asset manager (name withheld)
- Industry
- Asset management · Beacon
- Region
- Singapore / Asia-Pacific
- Solution
- PICOLens (AI governance)
- Frameworks
- NIST AI RMF, ISO 42001
Figures from the engagement record. Indicative and self-reported figures are labelled.
Situation
Portfolio managers and analysts were using generic AI tools with no shared framework, raising data-leakage and consistency concerns. MAS, SFC and SEC expectations demanded auditability and explainability.
Task
Give the whole investment team practical, responsible AI fluency. Define where AI fits in the workflow and what to build or buy. Write the policy and model-governance framework.
Action
- Aligned the CIO and COO on strategy and Responsible AI posture.
- Ran three tracks: AI for research, Responsible AI for investment management, and hands-on agents.
- Pair-built prompts and agent flows with analysts on real cases.
- Wrote the policy, model inventory and evaluation criteria.
Result
- A board-approved Responsible AI policy now runs across the firm.
- A shared prompt library and evaluation harness sit inside the research workflow.
- The investment committee funded a roadmap for three research agents.
Related case studies
All case studies ❯A 2030 digital banking blueprint and a coreless architecture the whole bank adopted
Audit-grade technology due diligence before the regulator’s second review
Bring us your hardest, highest-stakes AI decision.
A 60-minute working session, no obligation. Your data stays yours.