AI in Insurance: Get More from Insurtech Insights Asia

A big insurance conference isn’t valuable because it’s big. It’s valuable because it compresses a year of decision-making into two days—if you show up with a plan.

That’s why Insurtech Insights Asia (Hong Kong) matters to anyone working on AI in insurance right now. The agenda is broad, the hallway conversations are sharper than most webinars, and you’ll be surrounded by people who are actually implementing things: underwriting automation, claims triage, fraud detection, agent copilots, pricing modernization, and the unglamorous—but decisive—work of data governance.

The original announcement from The Digital Insurer (TDI) shared a practical perk: TDI members can use a 30% discount code (TDIASIA30) for the event. But the bigger story is how to turn an event like this into pipeline, partnerships, and a clearer AI roadmap for 2026.

Why AI in insurance is the real headline at ITI Asia

Answer first: If you attend Insurtech Insights Asia with an “AI-first lens,” you’ll spot which AI use cases are scaling, which are stuck in pilots, and what operating models insurers are using to ship results safely.

At the end of 2025, most insurers aren’t debating whether to use AI. They’re debating:

• Where AI creates measurable business value (loss ratio impact, expense ratio reduction, conversion lift, cycle time)
• How to deploy GenAI without a compliance headache (model risk, explainability, privacy, audit trails)
• How to staff it (central AI teams vs. product-led squads, vendor partnerships, “citizen developer” guardrails)

Events like Insurtech Insights Asia tend to surface the implementation details people don’t put in press releases: where the data breaks, how the workflow actually changes, and what metrics executives are demanding.

The four AI themes you should listen for

Answer first: The most useful conference content clusters around four themes: operations automation, decision intelligence, trust & governance, and distribution productivity.

1. Operations automation: Claims intake automation, document understanding, call summarization, straight-through processing, and case routing.
2. Decision intelligence: Underwriting risk selection, propensity models, pricing modernization, portfolio steering, and early warning for claims severity.
3. Trust & governance: AI model governance, human-in-the-loop design, bias controls, auditability, and third-party risk.
4. Distribution productivity: Agent/broker copilots, next-best-action, lead scoring, and hyper-personalized customer engagement.

If you hear a session talk about “AI transformation” without touching at least two of these, it’s probably more branding than substance.

Three reasons insurers should attend (beyond “networking”)

Answer first: The best ROI comes from (1) faster vendor due diligence, (2) stealing proven operating patterns, and (3) validating your 12-month AI roadmap.

1) Vendor and partner due diligence in days, not quarters

Procurement cycles can drag on for months. Conferences compress that timeline because you can do real-time comparisons—especially for tools that look similar on paper.

If you’re evaluating AI vendors for underwriting, claims automation, or fraud detection, you can quickly pressure-test:

• Data requirements (What do they need on day one? What do they claim they need “later”?)
• Integration reality (How do they fit into your policy admin system, claims system, CRM, or data lake?)
• Operational workload (Who labels data? Who monitors drift? Who handles exceptions?)
• Security and compliance posture (What’s logged? What’s retained? Where does data flow?)

I’ve found that the fastest way to spot fluff is to ask one question: “Show me the workflow for an exception.” AI demos are always perfect. Exceptions are where ROI dies.

2) Implementation patterns you can copy

Insurance leaders are converging on a few practical patterns for deploying AI responsibly:

• Human-in-the-loop by design for high-impact decisions (underwriting declines, fraud referrals, claim denials)
• Tiered automation (e.g., fully automated for low-risk segments, assisted for medium, specialist review for complex)
• Use-case scorecards that include both business impact and risk controls
• Model monitoring as an operational capability, not a data science afterthought

Events are one of the few places you can learn how others staffed these patterns and what they did when regulators, auditors, or legal teams pushed back.

3) Roadmap validation (and ruthless prioritization)

Everyone has a long AI wishlist. The smart move is to leave with a shorter list.

If you’re building an AI roadmap for 2026, your goal isn’t to collect ideas. It’s to answer:

• Which 2–3 AI use cases will move the needle in the next 90–180 days?
• What data blockers will stop us—and how do we remove them?
• What governance “non-negotiables” should we set before scaling GenAI?

A good conference forces honesty. If five peers say their GenAI assistant stalled because knowledge bases were messy, that’s your cue to fund content ops and data hygiene—not another chatbot pilot.

Where AI use cases are paying off in insurance right now

Answer first: The strongest near-term returns show up in claims cycle time reduction, contact center productivity, fraud triage, and underwriting decision support.

Here’s a practical map of AI in insurance use cases you’ll likely see discussed—plus what to ask to separate real deployments from prototypes.

Claims automation and claims triage

What’s working:

• Automated document ingestion (FNOL forms, invoices, medical documents)
• Summarizing adjuster notes and call transcripts
• Routing claims based on complexity and suspected fraud indicators

What to ask:

• What percentage of claims are auto-triaged today?
• How do they handle low-quality photos, incomplete forms, or contradictory data?
• What’s the measurable change in cycle time (days) and leakage?

Fraud detection through AI (and smarter investigations)

What’s working:

• Network analytics to identify organized fraud rings
• Risk scoring to prioritize SIU investigations
• Automated evidence packaging for investigators

What to ask:

• What is the false-positive rate, and who absorbs the cost of reviewing flags?
• How do they avoid bias in fraud scoring?
• How do they measure “investigation productivity” (cases per investigator, recovery per case)?

Underwriting automation and decision support

What’s working:

• Risk appetite guidance and next-best-questions for underwriters
• Data prefill from third-party sources to reduce manual input
• Portfolio-level insights to adjust rules and pricing

What to ask:

• Does the model explain why it recommends an action in a way underwriters trust?
• How do they manage model drift and changing risk profiles?
• What’s the escalation path when the model conflicts with underwriting judgment?

Customer engagement and agent copilot tools

What’s working:

• Drafting emails and policy explanations in plain language
• Call summarization and automated follow-ups
• Guided selling and cross-sell prompts grounded in policy context

What to ask:

• Is the assistant grounded in approved product language and updated documents?
• What prevents hallucinations from reaching customers?
• How do they log interactions for compliance and dispute resolution?

A two-day playbook: how to attend and come back with outcomes

Answer first: Set three outcomes, build a question bank, and schedule “decision meetings” during the event—not after.

If your goal is leads (or pipeline) for an AI program, a conference can be a goldmine. But it’s not automatic. Here’s a simple approach that works for insurers, brokers, and insurtech teams.

Before you go: define outcomes that fit on one page

Pick three outcomes, not ten:

1. One AI use case to pilot in Q1 2026 (with an owner and a success metric)
2. Two vendors/partners to shortlist (with a clear next step and timeline)
3. One operating model improvement (governance, monitoring, data access, compliance workflow)

Then write your “success metric” in plain terms, such as:

• Reduce average claims handling time by 15%
• Increase straight-through processing by 10 points
• Cut contact center wrap-up time by 30 seconds per call

During the event: use a question bank

Use the same questions repeatedly. Consistency makes comparisons fair.

• What’s live in production, and since when?
• What’s the integration effort in weeks, not months?
• What data do you require, and what do you generate?
• How do you handle exceptions and edge cases?
• What do you monitor weekly (drift, accuracy, cost, throughput)?
• What’s your governance model for GenAI outputs?

A useful rule: if a vendor can’t explain monitoring and exception handling clearly, you’re looking at a demo—not an operating solution.

After the event: run a 72-hour “conversion sprint”

Most conference momentum dies in the inbox. Don’t let it.

Within 72 hours, do three things:

1. Hold a 30-minute internal readout: what you learned, what changed, what you’re stopping.
2. Book two deep-dive sessions with shortlisted partners, with the right people in the room (IT, security, compliance, ops).
3. Write a one-page pilot charter: scope, data, workflow, owners, risks, KPI, go-live criteria.

If you do only this, the conference pays for itself.

Discount details (and how to use them strategically)

Answer first: If you’re already a TDI member, use the 30% discount code to lower the cost of attendance—and reinvest the savings in preparation and follow-through.

The TDI announcement highlighted a member benefit: 30% off registration with promo code TDIASIA30.

My take: the discount is nice, but the smarter move is what you do with the savings.

• Pay for one extra colleague from claims, underwriting, or compliance to attend with you
• Schedule vendor deep dives in advance so you’re not fighting calendars
• Build a shared notes doc with your question bank and decision criteria

Conferences are expensive when they become “inspiration trips.” They’re cost-effective when they become decision accelerators.

People also ask: quick answers for busy insurance leaders

Is Insurtech Insights Asia worth it for AI in insurance teams?

Yes—if you attend with a shortlist of AI use cases and a plan to validate them. The value is speed: you can compare approaches, vendors, and operating models in two days.

What should I focus on if I’m early in AI adoption?

Start with claims automation and contact center productivity. They’re easier to scope, deliver measurable savings, and build confidence for harder use cases like pricing and underwriting automation.

How do I keep GenAI compliant in insurance workflows?

Use grounding in approved sources, apply human review for high-impact outputs, log prompts/outputs for audit, and define escalation paths for exceptions.

What you do next matters more than what you hear

AI in insurance is moving from experiments to operations. The winners aren’t the companies with the most pilots—they’re the ones that operationalize two or three use cases and scale them without losing control.

If Insurtech Insights Asia is on your calendar (or should be), treat it like a working session for your 2026 AI roadmap. Walk in with a question bank, a shortlist, and a commitment to make decisions quickly when you get back.

And if you’re a TDI member, the TDIASIA30 discount is a simple nudge to attend. The real payoff is leaving Hong Kong with one clear answer: which AI initiative will you ship next—and what will you stop doing to make room for it?