AI agents for compliance teams in capital markets in Singapore

For Singapore fund managers and brokers, AI agents are most useful checking accredited investor evidence, matching the representatives register to who actually advises clients, reviewing cross-border marketing and triaging alerts. Compliance keeps every decision to report, onboard or restrict. MAS licensing rules, the FEAT principles, the Technology Risk Management Guidelines and the PDPA set the boundaries.

What does capital markets compliance involve in Singapore?

Singapore hosts a large number of fund managers, many serving clients from across Asia, along with brokers, private banks and family offices. Compliance in these firms is shaped by who the clients are and where they live.

Investor classification is central. Many products can only be offered to accredited or institutional investors, and individuals must opt in to accredited status after being told what protections they give up. Compliance checks that each client's classification is evidenced and that products sold match it.

Representatives must be notified to MAS before they conduct regulated activities, and the public register must match reality. Cross-border marketing to clients in other jurisdictions brings those countries' rules into play. Trade surveillance, AML checks on wealthy clients with complex structures, suspicious transaction reporting, and annual attestations to management complete the work.

Where do agents help?

1. Accredited investor evidence (depth 1). The agent checks each client file for the opt-in, the evidence of wealth or income, and the date of the last review, and lists clients whose evidence is stale or missing.
2. Representative register reconciliation (depth 1). The agent compares the firm's list of staff who speak to clients, meeting records and email activity with the representatives register, and flags staff who appear to be advising without notification.
3. Cross-border marketing review (depth 2). The agent reads materials and meeting notes for clients outside Singapore, notes the client's country, and drafts a checklist of that country's restrictions for compliance to confirm.
4. Source of wealth summaries (depth 2). For private clients with trusts and holding companies, the agent drafts a source of wealth narrative from documents provided, citing each one, for the AML analyst to test.
5. Surveillance alert triage (depth 2). The agent reconstructs order activity around price moves and drafts a disposition.

Representative reconciliation is a distinctively Singapore use. The register is public and the obligation is precise, so an agent that finds the junior analyst who has quietly started running client calls catches a real breach early.

What stays with compliance?

Reclassifying an investor, deciding that a staff member was conducting regulated activity without notification, reporting to MAS, filing suspicious transaction reports, and accepting a private client with a complex source of wealth all stay with named people. MAS expects senior managers to be clearly accountable for these areas, and its guidelines on individual accountability make that explicit.

Cross-border judgement stays human. Whether a conversation with a client in another country amounted to soliciting business there depends on facts and foreign law the agent cannot weigh with confidence. The agent drafts the checklist; compliance or counsel decides.

This reflects our general view that AI compliance agents are second line, never the last word. The agent extends coverage across every client file; it does not take the call.

Which rules apply in Singapore?

MAS licenses and supervises capital markets intermediaries, including fund managers and brokers, and sets conduct, business and AML/CFT requirements for them. Singapore's securities law prohibits market manipulation and insider trading and governs offers of investment products, including the accredited investor regime. MAS's guidelines on individual accountability and conduct expect clear responsibility for key functions.

For AI, three frameworks matter. MAS's FEAT principles (fairness, ethics, accountability, transparency) guide the use of AI and data analytics. MAS's Technology Risk Management Guidelines cover governance, testing and third-party risk for technology the firm relies on. IMDA's Model AI Governance Framework gives practical guidance on oversight and operations.

The Personal Data Protection Act 2012, administered by the PDPC, applies to client and staff personal data, including meeting notes and emails the agent reads. See our piece on the Singapore PDPA and AI agents and our overview of how AI agents are regulated in Singapore.

This is an explanation of the rules, not legal advice; confirm how they apply to you with counsel.

How do you start in the first 90 days?

Month one: run accredited investor evidence checks at depth 1 across the whole client book. This produces a clear, checkable list and often finds stale reviews the team already suspected.

Month two: add representative register reconciliation at depth 1. Agree with staff and the data protection officer which communications data the agent may read and for what purpose.

Month three: pilot source of wealth summaries at depth 2 on newly onboarded private clients, with every narrative tested by an AML analyst. Complete the technology risk review of the supplier before live use.