When Hasani Jess (pictured), chief technology officer at Aventum Group in London, describes the company's transformation program, he doesn't reach for modest language. "We're working on what we believe to be the most ambitious transformation program in our domain," he says. For a global insurance group operating across specialty lines in and beyond the London market, that ambition is not incidental - it is structural.
Aventum's technology overhaul is centred on a proprietary platform called ATOMX, built entirely in-house and designed to power the full lifecycle, from data ingestion and workflow management through to product distribution. According to Aventum, the platform is expected to save 1,000 staff hours monthly and reduce costs by more than US$2 million annually - backed by a US$12 million investment over two years announced in January 2025. The goal is not incremental improvement. It is, as Hasani puts it, to move from a successful specialty business to one that is "increasingly adding more lines of business through our growth plans, but doing so in a more efficient, more digitally native way."
The insurance technology market is full of point solutions - specialized tools that do one job well and depend on human effort to bridge the gap to the next system. Hasani calls this "human glue": the double-keying, the manual data transfers, the errors and omissions that accumulate wherever two platforms were never designed to speak to each other.
Rather than stitching those systems together, Aventum chose to build its own end-to-end solution. Off-the-shelf tools, Hasani argues, are typically designed to solve a specific point problem for a specific type of user. They weren't built with a holistic workflow in mind. For a multi-line group with ambitions to scale across its global broking and MGA operations- each with its own data requirements and workflow logic - that limitation becomes a ceiling. Hasani and his team wanted to build something with no ceiling. Readers tracking how AI is reshaping specialty insurance distribution can find ongoing coverage in Insurance Business America's dedicated technology section.
The platform is built and managed entirely by Aventum, with the company retaining all intellectual property. Development capacity is supplemented by what Hasani describes as augmentation partners - teams that embed within his organization rather than operating as external vendors. For AI model capability, Aventum works with OpenAI, Anthropic's Claude, Google's Gemini, and IBM, and is hosted on Microsoft Azure infrastructure.
Running a business across multiple specialty lines is conventionally viewed as a source of operational risk. Hasani sees it differently. The variety of data problems Aventum's team encounters - across different lines, different document types, different data structures - makes the platform's AI more capable, not less.
"The more variety of statements of value I get, the more my data science team can start to see how we need to approach processing those data files," Hasani explains. "Because we're getting to see a bigger range of complexity, we can bake those nuances into our AI and apply our training to a different set of problems." The result, he argues, is that training built to process one document type can be leveraged when categorizing email or ingesting data from entirely different sources.
It is a philosophy that parallels the logic of general AI itself. "General AI isn't focused just on one domain," he notes. "Its ability to understand football strategy somewhere along the line probably helps when you're doing architecture." For Aventum, diversity of problem statements isn't a complication to manage, it is an input that sharpens the platform's intelligence.
The technology is only one dimension of what Aventum is attempting. The other is cultural. Hasani is candid about the challenge of leading people through a transformation that has no natural precedent within the organization, or, he suggests, across much of the market.
"No one who works here has been through a transformation like this," he says. "There isn't that natural reference point." The response has been a sustained effort at what he calls vision casting, communicating not just what the platform will look like when it is finished but providing enough tangible progress along the way to sustain belief. "Vision casting only lasts so long. People buy into the vision, but then they need to see the hallway. The kitchen isn't ready yet, but they need to see something more."
That pressure for visible progress, in a domain as regulated and operationally complex as specialty insurance, is significant. Speed of delivery matters for credibility. Hasani acknowledges the tension directly: "Moving really, really fast in a domain as robust and regulated as insurance brings its own challenges."
The internal product organization Aventum has built follows Agile and product management principles, with dedicated product managers working directly with internal business users as customers. Feedback loops are real and continuous. "Product managers will be working with them after releasing a new feature and will be getting very vociferous feedback," Hasani says. "Whether they're happy, whether they'd ideally like some tweaks, or what the next thing on the roadmap should be."
The measure of success is not a completed platform but a live, evolving product that scales ahead of the business it serves. Whether that ambition will set a new benchmark for technology in the London market is a question the industry is beginning to ask.
