Insurers have spent the past two years racing to adopt AI. But according to Duck Creek's General Manager of Underwriting Andy Moss, the technology itself isn't what's giving market leaders an edge. And as a recipient of Insurance Business’s Global 5-Star Technology and Software Providers 2026 under subsidiary Send Technology, he should know.
“The real opportunity this year isn’t the technology on its own, it’s the redesign of the underwriting workflow so that modern technology can do its job,” he told IB. “Over the last decade, workbenches helped underwriters centralise information and reduce administrative burden, but the industry has now entered a new phase. AI, agentic automation, and richer data sources are advancing so quickly that traditional workflow simply can’t absorb them without rethinking how work moves from submission to bind.
“What we’re seeing across the market is that the insurers who focus on workflow transformation are pulling ahead. They’re moving away from document led processes and toward orchestrated underwriting environments where submissions are captured once, enriched automatically, routed intelligently, and governed consistently. This shift allows AI agents to support underwriters at scale by reading documents, extracting data, preparing pricing context, and surfacing insights, allowing underwriters to stay focused on judgment, negotiation, and portfolio strategy.”
And the data certainly chimes with Moss’s assertions here. According to data from McKinsey, AI-driven analytics can cut turnaround times for complex policies by up to 80% and boost agent sales success rates by 10% to 20%, with AI platforms even reducing processing costs by up to 30%.
As Moss went on to tell IB, the opportunity is not AI in isolation. Rather it’s AI working inside a redesigned, orchestrated underwriting flow connecting intake, triage, enrichment, pricing, referrals, and bind into one coherent experience.
“[Because] when insurers modernise the flow, the technology finally has room to deliver the speed, consistency, and precision the market now demands,” added Moss.
That market demand is really driving this digital transformation, with firms trying to outpace each other in the great AI arms race. Worldwide adoption of AI has surged this year, with 91% of businesses using AI in at least one capacity, global spending reaching $301 billion, and worldwide end-user spending on AI models projected to hit $64 billion.
But buying the tech is the easy part, the real difficulty comes in the orchestration stage. As Moss told IB, orchestration has become essential because underwriting has now outgrown the patchwork of systems insurers have relied on for years.
“Even after significant digital investment, many teams still operate with siloed data, manual handoffs, and fragmented submission intake,” he explained. “Underwriters spend too much time navigating email, spreadsheets, portals, and legacy systems that were never designed to work together.
“As AI adoption accelerates, this fragmentation becomes even more limiting. Orchestration solves the core problem by acting as the operating layer coordinating submissions, documents, data sources, pricing tools, appetite rules, and AI agents. Instead of forcing underwriters to stitch together their own workflow, orchestration creates a single governed environment where work moves automatically, decisions are consistent, and data is captured once and reused throughout the lifecycle.
“In short, orchestration has become a priority. Underwriters don’t need more tools, they need one connected system where all the tools, data, and decisions work together.”
Even with AI becoming more sophisticated, it still has the ability to make mistakes - or hallucinations - when left unchecked. This is where the human-in-the-loop concept comes into play. According to recent studies, eight in 10 business leaders say that human-in-the-loop in regards to AI is important for their organisation, with a further study finding that 90% of clients or consumers are more likely to trust a company that uses human-in-the-loop.
For Moss, he told IB that the balance between automation and human expertise comes from recognising that underwriting is fundamentally a judgment‑driven discipline.
“AI is most valuable when it strengthens, not replaces, judgment,” he explained. “Insurers are moving into an era where AI agents can read submissions, extract and enrich data, identify appetite fit, flag exclusions, and even prepare draft rationales. These capabilities only create value when they operate inside a controlled, orchestrated workflow keeping underwriters firmly in the decision‑making seat.
“We’re seeing underwriters benefit enormously when AI handles the operational load. Tasks that once consumed hours such as sorting inboxes, rekeying information, hunting for documents, assembling pricing context, can now be completed in seconds. This frees underwriters to focus on work to move the business forward from evaluating complex risks, negotiating with brokers, shaping portfolios, and applying their experience. It’s something no model can replicate.”
As Moss went on to tell IB, one critical ingredient here is governance, especially as agentic AI becomes more autonomous.
“Insurers need a workflow ensuring every action is auditable, explainable, and aligned with appetite, authority, and compliance rules. Orchestration provides structure because it coordinates AI agents, enforces thresholds, escalates exceptions, and ensures human‑in‑the‑loop validation wherever needed. This is especially important as insurers adopt multiple AI models and data sources - orchestration keeps them consistent, safe, and predictable.
“The insurers who strike the right balance won't treat AI as one thing. They'll know which decisions call for a co-pilot with a human firmly in the loop, and which can be confidently delegated to agents once that trust is earned. It's this mix, not a single operating mode, that defines mature adoption. In this model, trust isn't compromised, it's strengthened as decisions become faster, clearer, and more reliably governed.”
Looking ahead to the future, Moss believes that only insurers who look at AI as a tool for modernizing the overall underwriting processes will thrive, rather than those who see it as a quick fix.
As Moss went to explain, the insurers that truly modernise underwriting will rethink their operating model rather than layering new tools onto legacy processes, building connected, governed workflows to unify submission intake, triage, enrichment, pricing, referrals, bind, and renewal into a single orchestrated flow.
“This creates an underwriting environment where AI, data, documents, and decisions work together seamlessly,” he added. “Those who fall behind will continue adding point solutions which will increase complexity and fragmentation. They’ll still struggle with slow triage, inconsistent pricing, manual rekeying, and limited visibility into portfolio performance. The difference will be stark, with one group operating with speed, clarity, and discipline, while the other will continue fighting the same operational challenges with more technology but no improvement in flow.
“The winners will be the insurers that modernise the flow, not just the stack, creating a more profitable, more responsive, and more future‑ready underwriting operation.”
This article was created in partnership with Send Technology
