Insurance has spent the last few years investing heavily in AI, but many organizations are still struggling to turn that investment into meaningful operational gains. Pilots that look promising in isolated test settings can stall when reaching wider adoption, leaving expected efficiency gains unrealized and putting additional pressure on service teams.
An underrecognized part of the problem is timing. Many insurers have started reducing headcount before the automation expected to absorb that capacity is fully operational. According to recent Gartner data,
This creates a widening gap between intention and execution that puts insurance operations at risk. Preventing that gap depends on insurers steering away from automation-driven workforce cuts to instead prioritize operational readiness.
When automation isn't ready, the workload doesn't disappear
Instead of enabling true automation, many of today's AI investments are running into hurdles that directly affect service quality. When headcount is cut faster than workflows can be automated, backlogs quietly build until turnaround times stretch, service quality slips, and errors start showing up in the queues no one has time to double-check. The teams left behind after reductions are often responsible for absorbing the exceptions, judgment-heavy workflows and edge cases that still require human review. Rather than becoming an efficiency driver, these investments are silently becoming a hidden capacity deficit.
This is the core of the insurance math problem being faced by the broader industry: reducing headcount is treated as if it's the same action as automating a workflow. Yet, in practice, it's a bet that automation will catch up in time to cover the gap.
AI adoption is an operational challenge, not just a technology challenge
Part of the challenge impacting brokers, carriers and MGAs is that pricing, underwriting, claims and service still run on disconnected tools. Fragmented systems are only part of the challenge. AI also depends on the workflows, data standards and operational processes underneath those systems being structured well enough to support it.
Industry watchers have largely referred to this as an execution gap that widens even further when insurers invest in automation before their workforce and operating models are ready to support it.
With where we stand today, AI is only as good as the data it's given, and many legacy insurance systems weren't built with automation in mind. That underlying barrier is why many pilots that looked promising in an isolated test environment often break once they meet the complexity of real-world operational settings. The workflows and data underneath the technology are usually where it breaks down. They weren't ready to support what AI was asked to do.
Readiness is the real prerequisite for AI adoption
By prioritizing operational readiness, insurers are better able to ensure the groundwork automation depends on is laid from the start, not added after the fact. Until the workflows, data and staffing underneath these investments are prepared to support them, cutting people ahead of the automation curve pushes more strain onto the teams left behind.
When looking for ways to successfully adopt, a useful test for any organization looking towards these tools is to clearly evaluate who owns an exception when the AI can't resolve it, and how quickly that is expected to be handled. If the answer isn't easily identified, then the organization simply isn't ready to reduce headcount in its current state.
From working hands-on with carriers, brokers and MGAs, I've seen organizations sequence this backwards: reducing capacity first and expecting operational readiness to catch up later. The sequence needs to run the other way around to produce the returns AI investment is supposed to deliver.
Operational readiness should determine when and where capacity can be reduced, not be treated as an afterthought layered on once the reduction has already happened.
Building flexible capacity while automation matures
As demand soars and a retirement wave looms, I do believe insurers should keep chasing efficiency. I also believe what they've got backwards is the order of operations behind seeking it.
Insurers can build flexible operational capacity alongside their existing teams while readiness work is underway, giving organizations room to standardize workflows, strengthen processes and adopt technology without sacrificing throughput or service quality.
By doing so, these organizations are setting themselves up for success instead of assuming automation alone can carry the load on day one. That approach protects licensed and higher-complexity staff specifically, keeping their time focused on judgment-heavy decisions rather than absorbing overflow AI isn't yet equipped to handle.
This all boils down to a capacity problem. Insurers are treating it as a headcount problem. The goal should be to maintain throughput and service quality while the automation layer matures enough to deliver on what it was adopted to do. Flexible operational support can help absorb repeatable workflows, strengthen process consistency, and maintain quality control while automation capabilities continue to mature. This helps ensure throughput doesn't depend on the AI being ready before the people are.
To realize meaningful returns on AI investment, insurers need to sequence readiness ahead of reduction. The organizations that build the right workflows, data foundations and operational capacity first will be better positioned to adopt automation at scale without sacrificing throughput, quality or service.
