Insurance companies may be eager to adopt generative AI capabilities, but generic models often fail to grasp the industry’s complexities. That is the problem Roots set out to solve with InsurGPT, a domain-trained AI designed specifically for insurers.
Roots co-founder and chief executive Chaz Perera spent over 14 years within the insurance sector, as an underwriting and claims operations leader, eventually rising to the chief transformation officer level at AIG, a top-10 global commercial lines carrier.
During this time, Perera watched as insurers struggled to automate the immense amounts of unstructured data in nearly every insurance process. Traditional approaches typically involved mailroom document scanning – emails (often with several attachments), faxes, written notes, and even structured forms that didn’t fit existing automation templates well – and then passing them to a human expert to read, understand, and make decisions… dozens, even hundreds of times every day.
With few changes, this inefficient, time-intensive, and error-prone approach was insurance’s answer to its data challenges for the past half-century – and it was costing the sector billions of dollars yearly. Perera knew that by implementing AI capable of understanding unstructured data, insurance firms could move beyond the practices of the past.
“It is hard to solve that problem within the four walls of an insurance company,” Perera observes, “because there are a multitude of challenges that need to be solved, and this might not get priority. But we felt like if you could solve for unstructured data, you could substantively, move the needle across the industry. That’s why we started Roots, not to singularly help one company, but help all insurance companies at the same time.”
In the insurance business, profitability and growth largely boil down to accuracy in insurance risk assessment and policy pricing on the underwriting side and minimizing claim leakage at the other. Making good decisions within insurance depends on the ability to collate data from a wide range of sources, quickly. Roots empowers insurance professionals – underwriters, claims specialists, and brokers -, by delivering instant access to precisely the data they need to make the most effective decisions.
Since its founding in 2018, the company’s offerings have greatly expanded to enhance its intelligent automation capabilities, delivered through no-code, cloud-based AI Agents. At the heart of its technology is its advanced insurance LLM, InsurGPT™, which is built and trained by insurance professionals to understand the industry’s specialized language and intricate regulatory environment. Customers leveraging Roots can design, deploy and manage a team of AI Agents that support the entire insurance lifecycle, ultimately reducing the tedious workloads of staff.
Standing out from the crowd
A recent report from Fortune Business Insights estimated the global InsurTech market was valued at $15.56bn in 2024 and could reach $19bn in 2025. Within this boom market are a dizzying array of solutions for insurance firms to pick from, further complicating digital transformation for players in an industry that’s been traditionally lagged in tech adoption.
Several factors set Roots apart within its market. As firms race to implement modern generative AI and other tools, the transition to these technologies creates risks for a business that’s congenitally risk averse. “Hallucinations” and bias, often caused by model drift, the lack of transparency and explainability, which are inherent in AI solutions – particularly ones built on public AI models. Roots’ approach is focused on delivering trustworthy AI by building and training AI in-house on the industry’s most extensive corpus of non-private insurance data, and sharing its performance accuracy benchmarks publicly.
Perera also noted that Roots’ solutions weren’t simply created by engineers that “found” a problem. They were purpose-built by insurance experts with decades of first-hand experience struggling with insurance’s growing data challenge, and who wanted to make insurance work better for the more than 10 million people employed globally in insurance.
One distinction Perera makes to define Roots from others in the crowded insurance AI space is a holistic approach to digital transformation in insurance. “We’re not trying to solve like a sliver of a process or a sliver of a problem,” Perera adds. “We want to solve it, end to end because we fundamentally believe that’s the only way to get to value consistently.”
A new way forward
Roots entered 2025 with an extensive rebranding to better reflect the company during its next era of growth. Perera said, “This rebrand isn’t just about a new name—it’s about where we’re going and the technology we’re pioneering.”
A core part of the rebrand was to shift away from a focus on automation to communicate its wider AI capabilities. Perera added that, “the concept of automation was holding us back in people’s perception of Roots. The essence of our company is enabling humans to make good decisions and have all the information at their fingertips to make the right judgement calls. That’s not automation. That’s people at companies doing right by their customers, their colleagues, and their employer. To us, that’s the essence of a tree’s roots and that the roots of a company are its people.”
Generative AI purpose-built for insurance
AI is advancing at a breakneck pace. Only a few years ago the technology was used in data processing and other repetitive tasks Now, agentic AI and generative AI are expanding the potential use cases in insurance.
AI tools are becoming more cognitively “human.” While AI still needs oversight, it can process data and make decisions based on the instructions provided and even converse fluently with users. These burgeoning capabilities now allow the AI to control more of the workflow, supporting human experts throughout whole processes, rather than on one specific task.
“There is too much human involvement on work that most humans don’t like doing, activities related to processing forms, reviewing submissions, or identifying legal demands. It’s not interesting work — and definitely not the type of thing that gets you jazzed to get up in the morning and go to work. So, let’s take it off the human’s plate, and let’s get them focused on the things that they really care about.”
While insurers might be tempted by public generative AI models (e.g., ChatGPT, Google Gemini, et al), these prompt-driven products are not designed to meet the specific challenges of the sector.
Roots built InsurGPT™ specifically for insurance’s intensive demands. It doesn’t just read and respond; it’s trained to fully understand the role of an underwriter, claims adjustor or other insurance expert.
“It fully understands each of the different insurance products at the same level of complexity as a 20-year insurance veteran. It grasps the nuance of language, geography, and culture, because it’s trained on hundreds of millions of non-public insurance documents. That’s not something that Open AI or Anthropic and others can say relative to the insurance industry.”
The Roots proprietary AI model’s deep domain knowledge means insurers get accurate results 98% of the time. “You don’t have that level of certainty from other players,” Perera added. Ultimately, this allows insurers to reduce the amount of human oversight the technology needs, increasing the value of the AI tool.
However, if a business hesitant about trusting an AI model, the Roots Platform includes insurance professionals in the process with Human-in-the-Loop tools and customizable confidence scoring to determine how much (or how little) human oversight is desired.
Oversight and trust are only two factors complicating AI adoption. Many insurance firm’s tech stacks have complex legacy systems with technology spanning several decades, creating high potential for IT debt. Legacy tech drag is often a serious impediment to implementing InsurTech solutions. Roots’ AI is designed to adapt to its working environment. If a human can navigate through the legacy software through mouse clicks and keystrokes, their AI Agents can. This future-forward capability enables insurers to add Roots’ AI to workflows, without having to replace their entire tech infrastructure.
Growth plans
Looking beyond its rebrand, Roots is formulating an ambitious future growth strategy. At present, the company is focused on the US P&C market, with clients spanning the UK, Australia, New Zealand, and Singapore. It plans to expand into global P&C markets, as well as increase its presence in life and health insurance.
On the technology front, Roots is preparing to expand the number of use cases it supports. “When we think about our AI Agents, they support 25 different use cases across underwriting and claims. But there’s probably 100 use cases across just those two functional areas that we want to be able to help our customers to accomplish better, faster, and cheaper.”
In outlining the present state of his business and its future growth strategy to deliver global access to its innovative AI Agents, Perera, makes his case for why firms should work with Roots. “We are proven solution with 80+ AI Agents working at production scale. We’re not an experiment for the customer.”
“We’re an AI foundation for a company that is either early in their AI journey or deep into their AI journey but want to support multiple use cases. And then lastly, we provide a guarantee on our work. Nobody else does that.”
Roots was recently named in the eighth annual InsurTech100, a list that highlights the must-know companies in the sector. Read the full InsurTech100 list here.
Read the daily FinTech news
Copyright © 2025 FinTech Global
