Governed Wealth Intelligence
An agentic wealth-advisory layer governed end to end by SAFR components. The same agent, the same portfolio, and the same market signal across three clients produce three different outcomes: executed, queued for the relationship manager, or blocked. The mandate decides, not the model.
- Organisation
- OpenSAFR
- Context
- Built at SingHacks 2026 for the Julius Baer challenge track, "From Portfolio Monitoring to Intelligence: Reimagining Wealth Advisory."
- Components covered
- Governance EnvelopeAgent IdentityControls RepositoryDisposition EngineAudit Log
- Conformance level
- Not yet assessed. The conformance suite is planned.
[TODO: confirm the name of this implementation ("Governed Wealth Intelligence" is a working title).]
The problem it tackles
Relationship managers need proactive, explainable insight. The moment an agent can act, the question becomes whether it is permitted to act for this client, in this jurisdiction, under this mandate.
What SAFR looks like in private banking
Mandate type determines the outcome:
- Discretionary mandates allow execution.
- Advisory mandates produce proposals for the relationship manager.
- Execution-only mandates produce no recommendation.
Cross-border controls
Client residence, booking centre, and product registration are evaluated together across Singapore, Switzerland, Indonesia, and Hong Kong.
Evidence Quality and grounding
The Evidence Quality control feeds the Disposition Engine. The RM-facing grounding spans are an OpenSAFR extension, not part of the SAFR whitepaper.
Scenario cast
The same agent, the same portfolio, and the same market signal across three clients produce three different outcomes: executed, queued for the relationship manager, or blocked.
[TODO: summarise each client scenario at the level of "what outcome and why", without implementation detail.]
What we learned
Grounding catches unsupported claims, not flawed reasoning over true facts.
[TODO: add further honest limitations.]