As insurers race to move beyond AI pilots and into production, many are running into a familiar obstacle: they don’t have the people to do it.
Mayank Arora (pictured), global head coach for software engineering at FDM Group, said the AI skills crunch is hitting insurance harder than many sectors because of its legacy technology and regulatory burden.
“The insurance industry is really feeling the pressure when it comes to AI talent,” he said. Many carriers, he noted, still run on older core systems and have strong actuarial and risk functions but “often lack deep AI engineering and data science expertise.”
The gap becomes most visible when insurers try to turn proof‑of‑concepts into live systems.
“Running a proof-of-concept is one thing, but scaling to production requires specialized skills in areas like MLOps [machine learning operations], data engineering, and compliance – and those roles are hard to find,” Arora told Insurance Business. Strict explainability and governance requirements then add another layer of complexity that many teams are not staffed to handle.
Pilots typically run on curated datasets in controlled environments. Day‑to‑day operations look very different.
You really see the gap when insurers try to scale, Arora said. “Pilots usually run on clean, curated datasets, but real-world data is messy and siloed, so advanced data engineering becomes critical.”
He added that deploying and monitoring models in production demands MLOps skills that are still rare inside traditional carriers. Even when the technical work is done, AI literacy across business units can lag, slowing adoption and reducing impact.
Governance and explainability are another hurdle; few teams have experience building AI systems that meet regulatory and ethical standards, he said. “Finally, [it’s] change management – business units often lack AI literacy, so adoption stalls without embedded experts to guide the process.”
The supply‑side problem is compounded by intense competition for experienced practitioners.
“Honestly, it’s both,” Arora said when asked whether scarcity or cost is the bigger issue.
He added that senior AI engineers with domain knowledge in insurance are rare, and when you do find them, they’re expensive. Tech giants and startups offer higher pay and more exciting projects, so insurers struggle to compete, Arora said.
Even when carriers do manage to hire, he added, keeping those experts engaged inside conservative, highly regulated organizations is not straightforward. “Retention is a real challenge,” he said.
One strategy some firms are pursuing is to build AI capability from the ground up by embedding junior talent alongside experienced coaches inside carrier teams for extended periods, rather than relying solely on senior external hires.
Arora said his own work focuses on young professionals “who are passionate about AI and eager to build a career in this space,” paired with seasoned AI coaches who provide “hands-on guidance, best practices, and domain-specific knowledge” while working side-by-side with insurer teams.
For that kind of model, the biggest friction point is not technical – it is regulatory and cultural.
“The biggest challenge we face is regulatory compliance, governance, and change management,” Arora said. Younger engineers are “excited to explore AI,” he added, and “sometimes see compliance requirements as slowing them down.”
That is where he believes closer collaboration with insurers’ own risk and compliance experts is essential: carriers bring deep knowledge of frameworks and standards; external talent brings speed and experimentation.
“Introducing AI into a risk-averse industry requires cultural shifts, stakeholder buy-in, and clear communication,” he said.
Ultimately, Arora measures success less by the number of pilots launched and more by the capability that remains inside the organization once external support steps back.
He said domain knowledge and hands-on experience are critical to building that durability. High retention rates, he added, are one of the clearest signals that the model is working – when junior professionals convert into full-time roles and continue developing inside the insurance sector rather than leaving for tech.
For insurers, the challenge is not simply deploying AI tools, but embedding the skills required to maintain and govern them over time.
