Ravi Shankar (pictured), assistant vice president, solutions architect at Mosaic Insurance, did not set out to work in insurance. He said that after graduating, he “never thought” he would enter the sector, but that changed once he was exposed to it. He now describes insurance as central to his professional identity and says the attraction was both commercial and human: the industry, in his words, “touches everybody’s life” and matters most when clients face difficult moments.
His career took him across multiple markets and segments before settling in London. He worked in US insurance, spent time in Singapore and Malaysia, and ultimately chose London because of its strength in commercial and specialty business and its record of producing innovative products.
Asked what digital transformation means at Mosaic, Shankar set out a notably crisp framework. The company’s approach rests on three pillars: a data-first model, AI-led automation and API-native architecture. Those principles, he said, are not abstract design preferences but operating disciplines intended to improve the speed and quality of underwriting decisions.
He pointed to a concrete gain in cyber quoting. A process that previously took two to three days has, he said, been reduced to three or four hours by combining better data, API-led connectivity and selective use of AI.
At the centre of Mosaic’s current programme is an underwriting workbench, which Shankar described as a core initiative. Its purpose is to put the underwriter “to the forefront” of how the business captures and uses information, while connecting internal and third-party systems in a more structured way.
The workbench is also intended to do more than centralise workflow. Shankar said it can extract complex information for complex products, draw data from core and third-party systems via APIs and feed that information into Mosaic’s data lake so it can be analysed immediately and more meaningfully.
For Shankar, data difficulty in specialty insurance falls into two broad camps. The first is fragmentation: information sits across exposure systems, underwriting systems and other sources, and combining it into something useful remains harder than many outside the industry assume. He said that, despite progress on standardisation, stitching together data from different systems still creates significant challenges.
The second problem is data generated by people. In specialty lines, underwriters must interpret complex submissions and proposals, and that leaves room for error at the point of entry. Shankar said Mosaic is focusing on identifying these issues earlier and is deploying agentic tools to detect abnormalities before poor-quality data spreads through the system.
Shankar’s account of AI was striking for its restraint. Rather than presenting artificial intelligence as a sweeping cure-all, he said Mosaic prefers “small steps” and targeted use cases tied to a broader strategy. The company, he said, looks for specific operational problems and applies AI where it can solve them credibly.
He cited transactional liability as one example. Because the line involves large volumes of documentation, Mosaic has introduced an AI tool that extracts relevant information from those documents and surfaces it for the underwriter. Shankar said the system does so “with 90% efficiencies”.
Mosaic’s technology build is not, Shankar made clear, an exercise in reinventing what others have already proved. He said the insurer follows a philosophy of “partnership with the purpose”, preferring to assess proven market capabilities and then extend them, rather than build every component itself.
That partner selection process is tied to use-case validation. Shankar said Mosaic starts with a clear understanding of what it wants to achieve, then examines whether a potential partner has already done something meaningful in that area. If so, the task becomes one of scaling and adaptation rather than greenfield development. He added that partners also contribute training and experience drawn from work with other clients.
If there is a recurring theme in Shankar’s remarks, it is that transformation is rarely derailed by ambition alone; it is slowed by the need to align human adoption with technical deployment. He said Mosaic uses a “vanguard” group to test new tools before broader release, allowing the business to learn from early users before rolling changes out more widely.
The deeper challenge, he said elsewhere, is matching the pace of technological change with the pace at which teams can absorb it. Were he to begin again, he would spend more time with employees earlier in the process, giving them more information, involving them more directly and preparing them sooner for what is coming.
He was, however, notably relaxed about fears that AI might trigger a backlash from staff. Mosaic, he said, has a culture that is open to experimentation, provided people understand the point of the change and how it improves their work.
Shankar sees the next three to five years as decisive for the market. AI, he said, has moved beyond the proof-of-concept stage and is already affecting day-to-day insurance operations. At the same time, Mosaic is investing heavily in its data infrastructure, reflecting his view that insurance depends on “the right data one time every time”.
He also expects IoT and real-time information flows to become more influential. The direction of travel, he suggested, is towards better current data, stronger AI-assisted triage and risk assessment, and greater machine support for underwriters at the front end of the process.
Away from underwriting workbenches and API architecture, Shankar’s life is resolutely domestic. He said he is first a father and husband, with a five-year-old daughter and a 12-year-old son, and that family life can at times feel “even harder than digital transformation”.
He spends much of his time outside work with his wife and children and also keeps to gym sessions and long-distance running. In his own telling, he is “a big family man”, though one who also values the discipline and camaraderie of exercise.
