Xceedance said today that AI orchestration is producing measurable efficiency gains in live insurance operations, delivered through Orchestrated Intelligence, its operating model, and IAN, its agent studio for insurance.

In agency and broker operations, processes moved onto IAN delivered around 35% efficiency improvement from the point of go-live. As these implementations have matured, some are now approaching 50%. Xceedance is seeing the same pattern in underwriting support for excess and surplus lines carriers. The company is also deploying AI at scale in claims third-party administration (TPA) operations, where the volume of repeatable work is large.

A significant portion of gains comes from Xceedance teammates continuously refining AI-enabled workflows — correcting output, resolving the exceptions AI cannot handle, and incorporating those learnings into workflows and agent configurations. As processes mature, a greater proportion of routine work can be completed automatically, allowing teams to focus on higher-value activities.

Xceedance has been deploying AI across its operations for over a year, and that work has changed how the company sells and prices its offering. It is moving away from effort-based contracts towards transaction- and outcome-based ones. Clients increasingly pay for a policy issued, a claim processed, or a submission cleared, rather than for hours worked. Xceedance carries the risk of making the work more efficient, and the client sees the benefit in the price.

Orchestrated Intelligence brings human judgment, AI and software together to the same piece of work. It is not a choice between the three. Every task gets its own mix, so the work is done correctly at the lowest cost for that unit of work. Determining the right balance is informed by the expertise of insurance professionals at Xceedance who design and oversee these workflows.

IAN is where the agents are built and managed. Its design principle is that agents should be bounded: each one does a narrow, well-defined insurance task, is registered and versioned, and can explain its decisions. Bounding also keeps token costs down in deployed processes, because each agent uses only the context, prompts and model calls its task needs, so costs stay predictable.

Together, Orchestrated Intelligence and IAN have produced a catalog of more than 120 insurance agents spanning claims, underwriting, catastrophe modeling, finance and accounting, and data operations. Each agent was built for a single insurance micro task, such as classifying a submission, extracting fields from a loss run, or reconciling financial data. The catalog is available off-the-shelf, so most clients can start with agents already running on live volumes. It continues to expand as Xceedance applies IAN across additional insurance processes and workflows.

“The hard part in insurance is not building an agent. It is knowing what the agent should do, and then running hundreds of them without losing control,” said Arun Balakrishnan, Group Chairman and Chief Executive Officer at Xceedance. “That is what we built IAN for. We can design a bounded agent for a specific insurance task, put it into live work, and manage it at scale, in weeks rather than quarters. And we know what to point it at, because we have spent thirteen years doing this work ourselves. Orchestrated Intelligence is how it comes together — our teammates, agent studio and software, delivering the outcome on the same piece of work.”

About Xceedance

Since 2013, Xceedance has worked exclusively in property and casualty insurance, helping insurance organizations transform how work gets done through Orchestrated Intelligence, the coordinated application of human judgment, AI, and software. With 5,500+ teammates across the Americas, EMEA and APAC, we support 350+ clients, including commercial, personal and specialty lines re/insurers, mutuals, program administrators, brokers and agents, and Lloyd’s of London entities.

For more information, visit www.xceedance.com.

Media gallery

About The Author
Business Wire