How AI is driving efficiency, fraud detection, and customer-first outcomes in insurance.
Roi, please tell us a bit about yourself and Sprout.ai?
I’m a trained software engineer with a background working in tech start-ups and global companies including Tractable and Nice Systems. I joined Sprout.ai as CEO in 2022.
Sprout.ai is an award-winning technology company focused on transforming the insurance claims process. With ground-breaking AI products, the platform empowers insurers to make claims handling faster, easier, and more accurate, so claims handlers can focus more on providing empathy and support to customers.
Typically claims processes can take weeks or even months, but Sprout.ai uses AI technology to reduce this to near real time by eliminating error prone, manual processes that slow down the claims process. The technology is also effective at highlighting fraudulent claims, something that is particularly valuable now as The Coalition Against Insurance Fraud recently estimated that insurance fraud costs the United States an incredible $308 billion a year.
What are some of the main challenges facing the insurance sector at the moment?
It’s a challenging time for the insurance sector right now, with significant levels of claims inflation driven by a number of interconnected factors such as inflation, litigation, cost of repair or replacement and growing volumes of fraud. The findings from the Reinsurance Group of America (RGA) 2024 Global Claims Fraud Survey highlighted that 74% of respondents indicated the number of fraud cases is either holding steady or increasing compared to previous years.
All this means that claims are a growing cost centre for insurers and they need to balance this with adapting pricing, underwriting and finding efficiencies. In the US insurance premiums have been steadily rising across many areas and in March 2024 official headline US motor vehicle consumer price index (CPI) inflation was reported to be 22.2%, its highest level since 1976. The role of a leader managing the claims team is to find ways to improve efficiency and reduce cost, so – insurers are looking to their own internal processes to do this.
The industry is also experiencing talent shortages, partly due to an aging workforce and a battle to attract younger employees. This, coupled with rising expectations from customers about the speed, efficiency and quality of customer service, leaves insurers facing a struggle to protect their reputation and retain satisfied customers – which is where AI is offering a new opportunity.
Can you tell us a bit about how AI (and specifically Sprout.ai) can help address some of these issues?
In this context, AI is not merely an efficient tool but a necessity for tackling some of the greatest challenges the insurance industry faces. It enables insurers to process claims more swiftly and accurately, detect fraudulent activities and provide a better customer experience – something that may be contradictory to many people’s perceptions of AI.
AI technology is not just about replacing human tasks, but supporting human expertise. By automating routine processes, AI allows insurers to focus on complex cases that require empathy and nuanced judgment. This can ultimately lead to greater customer satisfaction as well as greater job satisfaction for claims handlers.
In the fight against fraud, AI can help insurers process and verify claims data with speed and precision, catching obvious discrepancies but also identifying subtle, emerging patterns of fraud. Taking on new technology will help insurers stay ahead of fraudsters, protect their financial health, and keep premiums low for honest customers – all of which is essential to ensure a fair and sustainable insurance market.
Finally, at a time when the industry is facing talent shortages, AI is able to help streamline workloads and support with efficiencies. It could be argued that AI technology such as that employed by Sprout.ai is the only way the insurance sector can successfully navigate these collective challenges.
What makes Sprout.ai different from the other companies in the same space/domain?
Sprout.ai acts as the “Brain behind the scene” and connects to the existing systems and processes insurers have. One of our biggest points of difference is the ease of integration and the fact we are dedicated and focus on insurance claims. We can drive value to our customers and within just 16 weeks. We have dedicated AI Models and APIs to the insurance claims world and we leverage synthetic data and proprietary deep-learning AI models to enable this. This means Sprout.ai’s technology can be integrated quickly without minimum training data – otherwise known as a ‘low’ to ‘no-data’ environment. This supports the information security needs of major insurance companies, where extracting training data can be a complex and expensive process.
Our technology also works across multiple insurance lines of business and multiple languages, so our customers, who are tier 1 global insurers, can use us in different areas of their business and different countries. We focus on providing a comprehensive solution to enhance the claim process, all the way from accurate data extraction through to automated policy coverage check, fraud flagging and providing a decision or recommendation on how to settle the claims. The AI models are capable of processing claims documentation in many languages, including character-based languages like Japanese and Chinese, which are the most difficult for AI systems to understand. Even operating in character-based languages, this documentation is processed at an average accuracy rate of 99%.
Our team includes ex-claims handlers and insurance professionals with first-hand experience of the industry’s challenges, which is why we feel we are able to offer such a unique and competitive product to the market.
Why do you think more insurance companies haven’t been quicker to embrace AI tech across all potential elements of the business – what is holding them back?
The insurance sector is, in general, a fairly traditional and risk-averse one which is partly why it has been slower than other industries to really embrace AI technology and partner with new, innovative tech providers. But, unlike other industries, insurance faces distinct technical hurdles in AI adoption. Legacy systems, some supporting policies that are decades old, present significant challenges. It’s important to remember we are an ageing population, which means some insurance claims will be based on policies and documents that are 60, even 80 years old! There are also some concerns within insurers about personal data, for example in health insurance claims where there may be large quantities of sensitive information to process.
Then there are the regulatory challenges to consider. Insurers have to ensure robust oversight of AI, akin to the governance applied to human operations. This includes tight compliance with a wide variety of regulations such as the UK’s Consumer Duty, GDPR, and HIPAA in the United States, all of which mandate stringent customer care and data protection standards.
As always, internal barriers such as entrenched cultures, outdated systems, and leadership mindsets can impede progress. Successful AI adoption requires clear goal-setting, training, change management, and the integration of best-in-class solutions. It’s essential to define expected business outcomes to avoid pilot projects that fail to scale.
For any insurers who are cautious about investing in technology and automation, what would you say to them? What’s your advice?
Insurers should start by deciding what the business problems are that they are seeking to solve and how they would measure success. By identifying some clear KPIs around for example percentage of STP (Straight Through Processing – a measure of the volume of claims processed end-to-end without manual intervention) turn around time for claims and customer satisfaction scores, businesses can align a Proof of Concept (PoC) or pilot with these to more clearly determine what works and what doesn’t – allowing them to refine accordingly before making any decisions about broader implementation.
Insurers should avoid the potential temptation to start too small with AI implementation projects. This type of digital transformation within a business requires bold moves and if you focus on only a minor area of the business with low risk, the reward may also be low and there is unlikely to be enough data to show progress.
When it comes to introducing new technology and AI driven systems into existing teams, it’s important for insurers to identify ‘change agents’ within the team that can act as internal champions to help drive adoption and reduce resistance to change. In a piece of research we conducted last year, claims handlers themselves told us that technology would greatly improve their role, with more than half (55%) stating that they wanted more data and insights tools to support them with their job. They were particularly keen for tools that could support them with the most tedious elements of their role which were identified as reviewing and processing documents (55%) and data entry and update (40%) suggesting that AI technology could be welcomed by many if its benefits in these areas are clearly communicated.
To reduce friction during adoption of new AI tools insurers should look for solutions that can integrate well with their existing systems and that are aligned with their current integration and data approach. By deciding where their competitive advantage lies and the core competencies of the individual insurance provider, companies can seek to partner with the right people to help them build their AI capabilities in this space rather than looking to buy solutions from the outset.
What regulatory shifts are you seeing globally that could significantly impact AI adoption in insurance?
I’ve already mentioned some of the global regulations the industry faces such as the UK’s Consumer Duty, GDPR, and HIPAA in the United States, which primarily focus on data protection and customer care standards. But one of the most interesting developments at the moment directly linked to AI is the legislation in the United States related to health insurance claims outcomes.
At the start of this year California enacted legislation to prohibit AI being solely used to make coverage claims decisions and to require physician oversight of the decision process. Arizona has since followed suit and this legislation has already been proposed by several other states and has the potential to influence legislation in other countries.
In the EU there is also the EU AI Act, which is the world’s first comprehensive AI law, adopted in June 2024. This law established a risk-based AI classification system with different risk levels meaning more or less AI compliance requirements. For example in the insurance industry, the Act lists the use of AI systems used for pricing in life and health insurance as high risk whereas systems used for the purpose of detecting fraud in financial services is considered to be lower risk.
Based on your experience, what’s the biggest misconception about AI adoption in highly regulated industries like insurance?
One of the most common misconceptions (not just in the insurance industry) is that AI technology will replace people, but in reality technology like Sprout.ai is designed to be a tool that can empower rather than replace workers. Human interaction and human empathy are core to customer interactions for insurers, which is why AI technology focuses on freeing up time by streamlining processes and reducing the administrative burden for claims handlers so that they can provide a better service to customers when needed.
There is also the issue of AI hallucinations that has been raised in recent years as the use of Generative AI has become more common. Hallucinations are instances in which an AI model generates content that’s plausible but actually is fictional or not based on real data. It can occur when the model extrapolates beyond its training data, and there are significant implications for insurers if these go undetected. While this is a real challenge at the moment as we are still in the relatively early stages of Gen AI adoption in the industry, the risk of hallucinations can be mitigated with the right strategies such as incorporating human oversight, ensuring robust training data for systems and utilising a combination of different AI models and cross verifying their outputs.
There is also a challenge around resource prioritisation. Business and Operations teams already stretched thin with customer service, backlog and ‘keeping-the-lights on’ project demands often lack the capacity to explore AI opportunities and there is a broad misconception that these will be extremely resource and time intensive to implement. Any system change obviously takes time to fully embed, but there are AI solutions such as ours that can be relatively simply and swiftly integrated into existing systems and processes.
Is there anything you think the insurance sector can learn from other highly regulated industries about how to successfully integrate AI solutions?
The insurance sector undoubtedly has much to learn from some of the failures and successes around AI adoption in other highly regulated industries such as finance, healthcare and legal services, but I also believe that these industries can gain valuable insights from what is happening within the insurance sector right now.
A common misconception is that AI adoption is primarily about efficiency. In reality, it’s about enhancing service quality, compliance and customer satisfaction as well, so anywhere where these are vital elements of a business there will be learnings from the insurance sector’s approach to utilising AI technology.
For industry leaders aiming to implement AI in heavily regulated environments, the key advice is to balance innovation with rigorous governance. Ensuring that AI systems are transparent, compliant and aligned with the organisation’s values is essential for building trust and achieving sustainable success.
What are your plans for expansion and growth? How do you see the company and the industry evolving in the upcoming years?
As a company we are really focused on working towards our overall vision of making every claim better. We already do this for over 12,000 people every day and we have a goal to grow exponentially until we reach our target of supporting one billion people around the world.
In terms of the evolution of the insurance industry and the use of AI, if we think about the ‘hype cycle’ around AI technology – we have moved beyond the peak of inflated expectations and the trough of disillusionment and are now entering the plateau of productivity where it is time for real results to be seen.
Looking ahead, I believe the tipping point for widespread AI adoption will come when the optimum balance is achieved between those who see only the risks and those who are energised and excited by the opportunities. As the adoption of AI grows, so will confidence and trust. During this period, some organisations may pull back, but others will refine their approaches and be the leaders of the change our industry needs.
The key to accelerating adoption is recognising AI for what it truly is: a tool to supplement and support, not replace, people and expertise. Practical deployments that deliver measurable benefits will help shift perceptions, enabling leaders and teams to see AI as a partner that improves outcomes and frees up energy to focus on higher-value activities. When this happens, insurance AI solutions will no longer be perceived as a threat that some see them as now but as an enabler of more high quality and consistent decisions and services, as well as a more positive work environment. Ultimately, the next ten years will determine which organisations have the purpose-driven leadership and determination to shape the future insurance success stories and which will become the brands of the past.
A quote or advice from the author : “The insurance industry is undergoing a shift similar to the digital overhaul that banking experienced a decade ago. At this stage AI adoption has evolved beyond theoretical discussion and proof-of-concept pilots to real-world implementation, impacting every part of the value chain – from underwriting and claims processing to fraud detection and customer service. I believe that, with the right focus and collaboration, AI has the potential to transform not just processes within the insurance industry, but the industry itself.”
Roi Amir
Chief Executive Officer, Sprout.ai
Roi is an accomplished and results driven Enterprise Software Executive with a proven track record gained managing teams in start-ups and global companies. He has demonstrable international experience in delivering enterprise level B2B products in a variety of domains and technologies as well as working with fortune 500 & FTSE 100 clients.
Previously with Tractable, Roi was instrumental in scaling the customer base and revenue, working with Tier 1 insurers such as Tokio Marine in Japan and GEICO in the US and helping to secure Tractable’s Unicorn valuation in June 2021.
Prior to that, at Zift Solution, Roi was the Chief Customer Officer, responsible for all post sales aspects of the customer work, including Delivery, Professional Services, Customer Success and Support. Accountable for customer satisfaction and ARR renewal of more than $20m and working with enterprise customers such as Canon, Cisco, IBM, Sage and more.