From doing to approving: the new operating model for insurance work

Every insurance organization is built around a basic assumption: a trained human handles the file. Claims examiners handle claims. Underwriters handle submissions. Servicing reps handle inquiries. The org chart, the systems, the workflows, the hiring profile, the comp structure — all of it presumes that the unit of work is a human handling a file.

That assumption is becoming wrong. Not in some far-off future. In the current quarter, in carriers that have deployed agents, the unit of work is increasingly an agent handling a file with a human approving the result. The implications run further than most people are accounting for.

The shift, stated plainly

The work that defines insurance — intake, document review, evidence assembly, decision preparation, write-back to systems — is moving from human hands to agent hands. The judgment work — coverage interpretation, appetite calls, negotiation, complex reserve decisions, fraud investigation — stays with humans. The approval layer becomes the new center of the operating model.

This is not a productivity story. It is a structural change in what humans are paid for. The carriers that recognize the change early reorganize around it. The carriers that wait try to preserve the old org chart and run the new technology underneath it. The first group gets a multi-year head start. The second group is paying for both models at once.

What changes in the org chart

The most visible change is in spans of control. A claims supervisor who today supervises eight to twelve adjusters supervises a much larger team in an approval-centric model — because the team's role is decision review rather than file build, and the supervisor's coaching surface is different. The same dynamic applies in underwriting.

The less visible change is in the entry-level role. Junior claims and underwriting positions have historically been training grounds where new hires learned the craft by handling files. In an approval-centric model, the work that taught the craft is not on the junior's desk anymore. Carriers that do not redesign the training pathway lose the on-ramp.

• Spans of control widen because the approval workload per file is smaller.
• Junior roles compress; the carrier needs fewer of them but each one matures faster.
• Senior roles expand in scope because they own the quality of the agent output.
• Subject matter expertise becomes more concentrated and more important — the senior expert is the source of the rules the agent applies.
• Workforce planning becomes about depth of judgment rather than breadth of capacity.

Designing the approval layer

The approval layer is the part of the operating model carriers underinvest in. It is treated as a residual — whatever the human still does after the agent has done the rest. That framing is wrong. The approval layer is the part the carrier is most accountable for, because it is the part where the carrier's judgment shows up in the file.

A good approval layer has structure. Approval gates are defined by dollar threshold, by risk severity, by counterparty, by jurisdiction. The decision the approver makes is bounded and documented. The agent prepares the recommendation; the approver confirms, overrides, or escalates. The reasoning is captured. The next file is faster because the prior approvals were structured.

The capacity unlock

The capacity gain is the part everyone notices. A team that was handling 100 files a day handles many multiples of that. The headline number is impressive, and it is also a distraction. The strategic capacity unlock is not in handling more files. It is in handling files the carrier could not previously handle well.

• Catastrophe surge handled without overflow staffing.
• Long-horizon claims monitored continuously rather than at scheduled touchpoints.
• Subrogation pursued on every flagged file rather than the sample examiners reach.
• Renewal preparation completed in advance rather than in the final two weeks.
• Fraud detection extended to the full population at intake rather than to flagged files only.

Each of those is a workflow the carrier had reasons it could not run at the desired quality. The capacity unlock makes them runnable. The dollar value of these workflows, summed, is typically larger than the headcount savings — and is the part of the case that gets undersold in the budget conversation.

What this means for the cost line

Cost reduction is a side effect of the operating model shift. It is not the headline outcome. Carriers that pitch agent deployment as a cost-saving project create the wrong internal incentives and miss the larger lift. The right framing is loss ratio, retention, capacity, and customer experience, in that order. The expense ratio improves as a downstream effect.

What carriers who get this right do

1. Lead with workflows, not technology. The conversation starts with claims and underwriting leaders, not with IT. The technology is in service of the workflow.
2. Pick one workflow, ship it, measure it, expand. The carriers that try to deploy everywhere at once usually do not finish anywhere.
3. Design the approval layer deliberately. Approval gates, dollar thresholds, override paths, and audit trails are all decided up front, not improvised.
4. Restructure spans of control on a normal cadence. Reorgs are not announced; they happen as roles are backfilled with the new operating model in mind.
5. Treat the senior expert as the source of truth. The rules the agent applies are the rules the senior expert articulates. The agent is the mechanism that scales the expert.

What carriers who get this wrong do

1. Deploy the technology under the existing org chart and never restructure. The result is a faster version of the same workflow, not the operating model change.
2. Treat agent deployment as a single IT program with a single timeline. The operating model shift is not a project. It is a permanent change in how the work is organized.
3. Optimize for token cost or model selection. The cost of the technology is rounding error compared to the cost of the work. Optimizing for the rounding error misses the prize.
4. Pitch the case as headcount reduction. The case is loss ratio and capacity. Pitching it as headcount creates resistance and wins less of the prize.

The bottom line

Insurance work is moving from doing to approving. The shift is operational, organizational, and strategic, in roughly that order of visibility. Carriers that lead with workflow choices, design the approval layer deliberately, and restructure spans of control on a normal cadence get the lift. Carriers that wait and run the new technology under the old chart get the bill without the lift. The next quarter is a fine place to start. The quarter after that is too.