“We’re selling a promise,” said Arvind Kaushal (pictured), founder and CEO of Cogitate. “That promise is a product which is driven from the data.”
In a few words, Kaushal cut to the core of the insurance business. As customer needs evolve and industry-specific risk solutions grow in demand, carriers are being pushed to rebuild their data infrastructure from the ground up - not to digitize, but to modularize.
In Kaushal’s view, the age of siloed legacy systems is ending. Where once a carrier’s tech stack was rigid and single-line focused - “I’m a worker’s comp system only. Oh, I’m a personal lines only” - modern platforms are being built to be business-line agnostic. “It has evolved now,” he said. The shift, enabled by SaaS delivery models and API-first development, has made the foundations of the insurance stack more flexible - but that’s only the beginning.
The deeper transformation lies in how data itself is handled. “There’s a shift from relational databases to Cosmos DB and all which are more file-based systems,” Kaushal said. These newer formats offer more agility, which is critical for building real-time, modular solutions at scale. Today’s architecture incorporates data from legacy core systems, client-owned platforms, third-party feeds, and live sources - all flowing into centralized lakehouses that are now core to modern insurance infrastructure.
“These lakehouses are really very rich,” he said. “And some of that is also live feeds which are going to them.” These real-time inputs not only allow insurers to respond faster, they also power advanced models, including generative AI and large language models (LLMs). “All these data lakehouses are suiting the LLM AR models, which makes it extremely easy to both… they are very modular and very easy to scale,” he said.
The implications for product design and underwriting are significant. With cleaner, more accessible, and constantly updating data sources, insurers can more easily tailor offerings, automate risk assessments, and support rapid configuration of industry-specific products.
Despite his enthusiasm for the shift, Kaushal issued a warning to insurers tempted to leap into transformation initiatives without a grounded strategy. “in my opinion, AI, LLMs are the breakthroughs… just don’t rush to them for the sake of it,” he said. He urged insurers not to fall into the trap of deploying technology for its novelty, rather than its business value.
Too often, insurers mistake a flashy proof of concept for a scalable solution. “On a POC, proof of concept, things look great, but when the rubber meets the road, that is where the challenge is,” Kaushal said.
His advice: start small, stay grounded, and build upward. “Take a simple use case and make it production ready,” he said. “Once you are happy with that, productionize it, and then build on top of that. It’s like building a Lego house, one by one.”
Looking ahead, Kaushal sees a more profound shift than just new platforms or improved data management. “There’s a fundamental shift happening,” he said. “From systems of records and digitization to systems with embedded agents, AI agents and intelligence built in.”
This move from digitization to intelligence, he said, is what will unlock the next level of customer experience and operational efficiency. As insurers look to compete in an increasingly tailored and tech-savvy marketplace, the goal isn’t just to digitize outdated processes, but to reimagine them with embedded intelligence at the core.
“It will help us to help our customers, to elevate the user experience, the customer experience of their customers,” he said. “And definitely operational efficiencies are going to come. That’s the shift we are looking at.”
