(Jatin Atre at Insurity’s Excellence event in October 2025. Source: Insurity.)
When Insurity (Hartford) named Jatin Atre President earlier this year, the company framed the appointment as part of a larger investment in AI, product development and customer support. The appointment came as Insurity committed approximately $50 million to AI and R&D, planned to hire more than 100 AI and machine-learning specialists, and added roughly 80,000 hours of customer support capacity. The company characterized the investment as an effort to embed AI directly into core insurance workflows rather than treat it as a standalone capability.
That context helps explain the unusually direct statement Insurity later issued on AI and core-system economics. Quoting Atre and Sylvester Mathis, Chief Revenue Officer and Chief Insurance Officer, Insurity, the company challenged carriers to judge AI not by the sophistication of vendor terminology, but by whether it measurably reduces the cost and time required to launch, change and maintain insurance products. The statement was striking in part because it was not simply a vendor’s assertion of its own AI capabilities. It was an invitation for carriers to demand more from Insurity and from core-system providers generally.
Jatin Atre, President, Insurity.
For Atre, the argument begins with a simple premise: AI should change the economics of insurance technology. If AI is treated as another layer of software, services or consulting spend, he suggests, the industry will have missed the point.
“Nearly every time a new technology comes in, its promise is to reduce the ultimate cost,” Atre says. “But it is often captured by the people who provide it, like the technology partners and the system integrators, and they actually increase the cost.”
The question, in Atre’s view, is whether AI will allow carriers to change their expectations of the systems that support their business. That means asking not only whether a vendor has embedded AI in a policy, billing or claims application, but whether the technology reduces implementation effort, shortens product-launch timelines, lowers maintenance cost and gives carrier teams greater control over work that has historically required outside implementation resources.
Changing the Economics of Product Launch
Atre’s argument turns in large part on the role systems integrators (SIs) have played in core-system projects. Policy administration implementations have often depended on large external teams to translate filings, rates, rules, forms and underwriting logic into working system configurations. That work has been expensive, time-consuming and, at least historically, partly defensible because it required scarce technical capacity and specialized knowledge.
AI changes that calculation, Atre argues. If a system can ingest filed product material, generate workflows and produce much of the initial structure that previously had to be built manually, then carriers should expect a different economic model. The question is not whether outside expertise disappears altogether, but whether the most labor-intensive parts of product setup and ongoing change should still require the same level of external services.
The question Atre raises is whether carriers should continue accepting implementation economics that were formed in an earlier technology era. If AI can now perform a substantial portion of the configuration and product-setup work that once required large external teams, then the burden of proof shifts back to vendors and SIs: they need to show why the old cost and timeline assumptions should still apply.
Atre says Insurity is spending a significant share of its AI investment on the work of ingesting filings and related product documents, using that material to generate workflows and reducing the manual configuration effort associated with rates, rules and forms. Much of the work that once consumed 18 to 24 months of implementation effort, he argues, can now be compressed because AI can generate a substantial first version of what previously had to be built by hand.
“I just feel that AI needs to be able to solve that problem,” Atre says. “And that’s what we are solving.”
In that context, Atre’s critique of SI-heavy implementation models is less a complaint about services firms than a claim about technological inevitability. If AI can perform a substantial portion of the configuration work that once required scarce programming capacity, then carriers have reason to expect the economic benefit to flow to them.
From External Dependency to Carrier Control
Atre says the work most likely to shrink falls into two broad categories: initial system configuration and long-term maintenance. The first includes the work of translating product filings, rates, rules, forms and underwriting intent into system behavior. The second includes the ongoing changes carriers need to make as market conditions, appetite, pricing and underwriting rules evolve.
For carriers, the distinction matters because of the obvious strategic consequences of speed-to-market. A product opportunity may exist for a limited window, especially in commercial and specialty lines, where market conditions, exposure concentrations and underwriting appetite can shift quickly. If a carrier needs 18 months or more to stand up or materially change a product, Atre argues, the opportunity may already have been captured by a faster competitor.
That problem extends beyond initial product launch. Atre points to routine business decisions—such as adjusting pricing or underwriting rules within filed boundaries—as examples of work that should not require the same multi-party, SI-heavy process that has characterized many core-system environments.
“Why shouldn’t you be able to run your own business at a fast pace when you see opportunities?” Atre says.
Sylvester Mathis, Chief Revenue Officer and Chief Insurance Officer, Insurity.
That question reflects a larger shift in how Atre believes carriers should think about core systems. The goal is not merely to implement a system successfully and then maintain it through periodic vendor and SI intervention. The goal is to give carrier teams more direct control over the systems through which they express underwriting, pricing and product strategy.
For commercial and specialty insurers, Atre argues, that is where AI has the greatest potential value. The most important use case is not simply helping a user find a policy, answer a narrow question or complete a task inside an existing application. It is helping carriers shorten the distance between a market opportunity and a working product.
Beyond AI Assistants
Insurity’s statement distinguished its approach from AI assistants and micro-agents focused on narrow tasks, particularly in personal lines and claims. Atre says the company’s focus is on policy administration for commercial and specialty carriers, where product complexity makes the cost and timing of product setup especially consequential.
That does not mean Insurity dismisses task-level AI. Mathis noted in the company’s statement that Insurity has put AI capabilities into production across underwriting, policy and analytics through its Andromeda and Borealis software releases. Those capabilities include risk intelligence, catastrophe modeling, submission scoring, underwriting workflows that surface claims data in context and AI-enabled premium audit self-service.
However, Atre’s broader argument is that AI’s value in core systems should be judged at the level of business outcomes. For commercial and specialty carriers, that means asking whether AI helps decide which risks to write, how to price them in relation to portfolio and catastrophe exposure, and how to bring new products and programs to market at the pace required by current conditions.
That emphasis makes the company’s challenge to carriers more pointed. Atre is not asking carriers to admire AI as a feature set. He is asking them to measure whether AI changes the economic assumptions of policy administration.
A More Composable Core
Atre also sees AI as part of a broader architectural shift. He argues that policy administration systems are moving away from large, monolithic structures toward sets of microservices that can be turned on, composed and orchestrated according to the needs of different lines of business.
That does not mean the core-system provider becomes irrelevant. In Atre’s view, it raises the standard such providers must meet. If a carrier can choose among specialized services, each component of a core-system provider’s platform must stand on its own. A rating service, intake service, issuance capability or underwriting workflow must be strong enough to compete with best-of-breed alternatives, while still benefiting from integration with the broader platform.
“The future of policy administration systems will always be in a set of microservices that can be turned on and off as it’s needed for different types of businesses,” Atre says.
That view acknowledges the pressure on traditional core systems without concluding that carriers will want to assemble their operating environments from dozens of separate vendors. Atre says carriers will still care about integration, long-term support and care and feeding of the systems on which their business depends. The challenge for core-system providers is to deliver that coherence without forcing carriers into the old economics of large-scale implementation and change.
Workflows That Learn from Users
Atre extends the same logic to the user experience. For decades, he argues, core-system vendors have expected insurance professionals to adapt themselves to vendor-defined workflows. Users learned the system, attended training and followed processes designed by product and UX teams.
AI creates the possibility of reversing that relationship. Rather than requiring underwriters or other users to work the way the system dictates, Atre says systems should learn how users work and adapt workflows accordingly.
He compares the shift to the personalization of consumer technology. A smartphone allows each user to organize applications and interactions according to habit and preference. Atre says enterprise systems costing millions of dollars should be at least as adaptable to the users whose daily work they support.
For underwriters, that could mean showing the system how they process a particular type of endorsement, evaluate a submission or organize the information most relevant to a line of business. The system should be able to learn from those patterns and shape workflows around them.
That vision is still emerging, but it is consistent with Atre’s central argument: AI should reduce the distance between insurance expertise and system execution. Whether the issue is product setup, maintenance, configuration or daily workflow, Atre sees AI as a way to place more control in the hands of carrier teams.
A New Burden of Proof
Insurity’s statement was unusually polemical because it challenged carriers to ask harder questions of vendors at a moment when AI terminology has become widespread across the insurance technology market. But the underlying message was not simply that other vendors’ AI claims should be scrutinized. It was that carriers now have a reasonable basis to demand more from every core-system provider, including Insurity.
That demand can be stated in business terms: How much will AI reduce implementation and maintenance spend? How much will it shorten the time required to launch or materially change a complex commercial product? Can carrier teams configure and control more of the system themselves, or will they remain dependent on vendor and SI resources?
Atre’s answer is that AI should materially alter those answers. If it does not, then the industry risks using AI to decorate familiar economics rather than change them.
“We’re challenging the industry on this because we’ve done the work, and we know what’s possible,” Atre says. “By applying AI to insurance product setup and configuration work, carriers should be able to materially reduce the time, cost and effort required to implement and maintain core systems. If AI isn’t changing those economics, it’s just more technology spend dressed up as innovation.”