Human Oversight Remains Central as Singapore Insurers Adopt Agentic AI

Insurers in Singapore are beginning to deploy agentic artificial intelligence to automate underwriting, document verification, and customer servicing, but final decisions carrying financial and regulatory risk still require human sign-off, according to German IT consultancy valantic. Speaking at the Asian Banking & Finance and Insurance Asia Summit in Singapore, APAC Director Gonçalo Carvalho demonstrated how AI agents can now handle the front-to-back workflow of life insurance processing while keeping human underwriters firmly in control.

Carvalho showed a live insurance workflow where AI agents independently verified customer identification, proof of address, and medical information before routing applications for underwriting. The AI then generated risk assessments — in one demo producing a "68%" risk score — but the human underwriter retained the authority to override the recommendation and approve the application. "It's not on your capacity but on my capacity to approve," Carvalho said. The workflow also generated offer letters, customer emails, and audit logs tracking whether each action was taken by AI agents or human users.

The demonstration comes as valantic predicts a 2,500% increase in enterprise AI projects by 2028, even while 67% of IT leaders remain uncomfortable with current AI reliability. Agentic AI marks a shift from passive AI assistants that recommend actions to autonomous agents that execute parts of business processes independently. But Carvalho stressed that governance guardrails and audit trails are non-negotiable: "Are we building solutions that are safe, secure, scalable, and governed properly?"

Why it matters for Singapore: As a regional insurance hub, Singapore is uniquely positioned to shape how agentic AI is deployed across Asian financial services. The Monetary Authority of Singapore's parallel work on the SAFR framework for AI agent safeguards in finance signals that regulators are paying close attention. For Singapore-based insurers, getting the human-in-the-loop balance right isn't just good practice — it's increasingly likely to become a regulatory requirement. The valantic demo shows what compliant agentic AI looks like in practice, with audit trails that satisfy both risk committees and regulators.