Manulife opens AI Center of Excellence in Singapore to scale insurance innovation across Asia
Manulife has launched an Artificial Intelligence Center of Excellence (CoE) in Singapore to accelerate AI adoption across its insurance operations in Asia. The focus is practical: embed AI into underwriting, distribution, operations, and customer interactions to cut cycle times and deliver more personalized advice.
The CoE will operate under a clear framework that emphasizes transparency, security, and responsible use. Singapore was chosen for its strong digital infrastructure, mature innovation ecosystem, and its regulated, no-nonsense approach to AI deployment.
What this means for insurance teams
- Underwriting: Faster risk assessment with triage models, structured and unstructured data ingestion, and improved pricing consistency. Human review stays in the loop for complex and edge cases.
- Distribution: Next-best-action guidance for agents and bancassurance partners, compliant content generation, and sharper lead prioritization based on consented signals.
- Operations & Claims: Automation for intake, document processing, fraud flags, and first-notice-of-loss routing. Expect shorter resolution times and cleaner audit trails.
- Customer interactions: Context-aware assistance across channels, clear explanations of coverage, and improved servicing with guardrails to prevent hallucinations and bias.
Responsible AI, not AI theater
Manulife says the AI program will be built on principles that are easy to defend with regulators: explainability, data minimization, security by design, and continuous monitoring. That includes model risk management, fairness testing, and human oversight where it matters.
The insurer plans to work with industry partners and regulators, including through the Monetary Authority of Singapore's Pathfinder initiative, to help shape sector standards. For context on Singapore's supervisory stance, see the Monetary Authority of Singapore.
Why Singapore
Singapore offers the right mix: talent density, cloud connectivity, and a clear regulatory playbook. It's a predictable place to test, validate, and scale models across markets with different levels of digital maturity.
The government's pro-industry posture helps, too. See the Singapore Economic Development Board for broader context on the ecosystem.
Talent and partnerships
Manulife plans to grow its regional AI bench over the next three years across data science, AI governance, and engineering. Expect mixed squads: product owners, MLOps, risk, legal, and distribution leads working off the same backlog.
The learnings from Singapore will be rolled out across Manulife's Asian markets, creating reusable components-data pipelines, evaluation harnesses, and policy packs-so teams don't start from zero in each country.
How carriers can put this playbook to work
- Start where the friction is highest: Claims intake, underwriting triage, policy servicing. Time-to-value beats grand strategy documents.
- Lock down data foundations: Clear data contracts, lineage, PII handling, consent capture, and retention policies before model training.
- Stand up model risk management: Document intended use, testing protocols, bias checks, safety guardrails, and fallback procedures.
- Design for human-in-the-loop: Set confidence thresholds for automation vs. escalation. Make overrides simple and auditable.
- Integrate into workflows: Ship AI inside the tools people already use-policy admin, CRM, contact center-not as a new app to check.
- Measure what matters: Quote-to-bind time, loss ratio impact, claim cycle time, FNOL resolution, NPS/CSAT, and compliance findings.
- Train the frontline: Agents and adjusters need prompts, checklists, and "what good looks like" examples-not just a login.
Guardrails insurers should adopt from day one
- Data boundaries: Keep customer data in-region where required. Use synthetic data for early experiments.
- Content safety: Templates for disclosures, disclaimers, and source citations in all AI-generated communications.
- Vendor posture: DPAs, SOC2/ISO attestations, model update notifications, and exit plans to avoid lock-in.
- Continuous evaluation: Regression tests on every model update; monitor drift and re-certify high-impact models.
What good rollout looks like
Pick two to three use cases with clear owners and simple KPIs. Build a shared evaluation harness, ship to a small cohort, measure lift, then scale.
Create a central pattern library-prompts, guardrails, red-team scenarios, and approval checklists-so every market benefits from the same hard-won lessons.
If you're building capability in-house
Upskilling is a force multiplier. For structured learning by role, you can explore curated tracks here: AI courses by job function. Keep the focus on applied skills: data quality, evaluation, workflow design, and compliance.
The bottom line
Manulife's AI hub in Singapore signals a practical shift in how insurers deliver value-faster decisions, cleaner workflows, and advice that fits each customer. The carriers that win will pair speed with discipline: real use cases, measurable outcomes, and responsible controls baked in from day one.
Your membership also unlocks:
