Talcott Financial Group entered the retail annuity market in January with three new products and a technology platform connecting 12 different systems. The company had spent the previous year preparing for the launch, compressing work that its technology chief said would ordinarily take two to two-and-a-half years.
The deadline was met, but the work was far from finished.
“They didn’t get a break,” Samir Srivastava, group chief information officer at Talcott Financial Group, told Newsweek, describing the employees responsible for building the platform. They moved directly into supporting it as the company began selling its new products.
Srivastava discussed the effort as part of the Seeking Value video series presented by Cognizant. He was joined by Haminder Gera, strategic business unit leader for insurance at Cognizant, which has worked with Talcott for about two decades. Gabriel Snyder, Newsweek’s executive editor, enterprise, moderated the conversation.
...
Talcott, a life insurance group backed by investment firm Sixth Street, had expanded through transactions involving existing blocks of insurance policies. Its retail expansion required the company to establish technology capable of selling and servicing newly issued annuities.
Insurers typically introduce a limited selection of products first, Srivastava said, keeping upfront spending manageable while establishing their sales operations and reputation.
“We wanted a full shelf of products, right away,” Srivastava said.
The company introduced three fixed annuities—one multiyear guaranteed annuity and two fixed indexed annuities—designed for customers with different retirement savings and income needs.
Srivastava noted that offering all three immediately would give Talcott a wider range of options for customers and opportunities to sell complementary products.
The company began selecting its technology and architecture near the end of 2024. That left 2025 for an intensive development effort aimed at having the platform operational by the start of 2026.
Meeting that date required coordinating technologies supporting a customer’s relationship with Talcott, from the initial sale through policy administration, servicing and eventual claims.
Gera identified integration as one of the most difficult parts of the project. The platform involved 12 systems, some supplied by outside software providers and others developed internally. All had to work together while meeting the quality and compliance requirements of a regulated financial business.
More broadly, Gera said AI is changing how quickly technology teams can develop, test and connect software. He described initiatives that previously took multiple years being completed in six to 12 months, with applications in development, quality assurance and user acceptance testing.
Talcott’s market entry has performed well, according to Srivastava. He said sales were exceeding the company’s initial expectations and nearing its more ambitious goals, although he did not disclose figures.
He credited the distribution team’s relationships, the product designs and the technology supporting the sales process.
The aggressive schedule also produced a lesson that Srivastava said he would apply differently in hindsight.
“We probably should have prepared for the stabilization period,” he said.
Srivastava said the team had to work through problems that emerged after launch while supporting live operations. There was little opportunity to recover from the intense development effort before taking responsibility for the finished platform.
Another major project brought Talcott and Cognizant together to transfer the administration of an existing block of insurance policies from another carrier onto Talcott’s systems.
In reinsurance arrangements, the company assuming financial obligations does not necessarily take over responsibility for servicing policyholders. That work often remains with the original insurer.
Talcott wanted to assume both responsibilities.
Srivastava said the decision reflected the company’s ambition to provide broader services to the insurance industry, including administering older policies. Such transfers can be difficult because customer records often reside in legacy systems, the data must be converted and regulators oversee changes affecting policyholders.
Talcott notified customers in advance that their policies would be administered by a different company.
“We were transparent about that because that’s just the right thing to do,” Srivastava said.
Policyholders still needed to complete transactions by phone or online with the same or better service, Srivastava said. That required clear responsibilities for extracting data from the original carrier’s systems and handing it over to Talcott.
Talcott and the original carrier established milestones and assigned responsibility for each stage, working with Cognizant to complete the transfer. Srivastava said the conversion finished under budget and slightly ahead of schedule.
“We have a working solution,” Gera said, adding that Cognizant is incorporating AI capabilities to make future conversions more effective.
Srivastava had traditionally favored buying and configuring commercial software. But existing tools for modeling how insurance liabilities respond to different economic conditions offered limited flexibility, and changing stress-test scenarios could mean lengthy waits for vendors. Talcott decided to try building its own tools using AI.
The company established a sandbox where actuaries and engineers work together, exploring ways to run more varied analyses and make adjustments faster.
“I don’t think we’ve proven things out yet, but it’s very promising,” Srivastava said.
The experiment raises another question: How can employees develop software quickly without bypassing the safeguards required in a regulated business?
Srivastava described an approach in which business specialists can experiment with and build applications, while IT remains responsible for preparing them for operational use.
Once a tool moves beyond the sandbox, it must meet appropriate standards for documentation, testing, auditability and regulatory examination, with more rigorous oversight for critical systems than smaller employee-built applications.
Talcott is still determining how to apply those distinctions as more employees become involved in developing their own tools.
Gera emphasized the importance of reliable data, system integrations and responsible AI as insurers develop new capabilities.
“We always think of technology as a productivity improvement,” he said. But he argued that companies should also pursue “process reimagination and process redesign” to improve the customer experience.
Srivastava sees a longer-term opportunity in using technology to tailor insurance services to individual customers.
“I think that’s still eluding a lot of [the] industry, certainly ours.”