My introduction to psychology came in high school when my best friend's mother earned a PhD in the field and began writing a syndicated newspaper column that led to a talk show on Pittsburgh radio. Sam snickered at the notion that anyone would take advice from his mom but wound up on her show as a reluctant guest some years later. The plan was for her to opine about one thing or another while drawing on his experiences as a scuba instructor at various Club Meds during his summers while in medical school. She invented a French last name for him to hide their connection, then ambushed him by quizzing him about his sex life with the tourists. 

I howled as I listened to him squirm, knowing just how much he wanted to scream, "Mom, leave me alone!" as I'd heard him do so many times over the years.

Suffice it to say that, in the face of Sam's mom's very public efforts to use her degree to build an audience, neither of us took the field of psychology seriously at the time. But I've learned a thing or two over the decades, now follow developments in the field and recently saw a study on the use of AI that, while plenty interesting on its own terms, could also have broad implications for how insurers can use chatbots to take better care of customers. 

An article in the MIT Technology Review describes an eight-week clinical trial with 210 participants, about half of whom had access to an AI called Therabot and half of whom participated in traditional psychotherapy. As reported in a journal of the New England Journal of Medicine, those suffering from depression experienced a 51% reduction in symptoms from interacting with the AI, which participants did on average through 10 messages a day. 

"Those with anxiety experienced a 31% reduction, and those at risk for eating disorders saw a 19% reduction in concerns about body image and weight," the article said. "These measurements are based on self-reporting through surveys, a method that’s not perfect but remains one of the best tools researchers have.

"These results, [the study's lead author] says, are about what one finds in randomized control trials of psychotherapy with 16 hours of human-provided treatment, but the Therabot trial accomplished it in about half the time. 'I’ve been working in digital therapeutics for a long time, and I’ve never seen levels of engagement that are prolonged and sustained at this level,' he says."

As you can imagine, the study comes with a bazillion caveats, and not just the usual ones about how clinical trials don't necessarily translate into the real world. The big issue is not just that it's hard to duplicate a therapist-patient relationship in software but that an AI that commits missteps in providing therapy can cause real harm.

A New York Times article warns about "two teenagers who had consulted with 'psychologists' on Character.AI, an app that allows users to create fictional A.I. characters or chat with characters created by others. In one case, a 14-year-old boy in Florida died by suicide after interacting with a character claiming to be a licensed therapist. In another, a 17-year-old boy with autism in Texas grew hostile and violent toward his parents during a period when he corresponded with a chatbot that claimed to be a psychologist. Both boys’ parents have filed lawsuits against the company."

But insurers can still take heart from the progress that Therabot represents and can use the study to start to see more clearly where the benefits will be for customer service in insurance. 

An attempt to build an AI therapist as an "expert system" flopped in the 1960s, but the Therabot researchers saw another opportunity as the language models that now underpin generative AI began to take shape. They experimented with what raw material to use to train the AI, starting with general mental-health conversations pulled from internet forums, then turning to thousands of hours of transcripts of real therapy sessions. The AI's advice based on those inputs was too bland, so the researchers created their own data based on best practices.

That's more or less what I see happening with customer service bots in insurance. There's lots of experimentation with what data to feed them, what things to allow them to say on which topics, and what tone they should adopt.

The fact that the Therabot researchers seem to have found a good balance for tone and information on an even trickier subject suggests that the work in insurance will pay off, too, and sooner rather than later. 

There's already been considerable progress. The Wall Street Journal recently ran a piece saying Allstate has found that its AI is more empathetic than a lot of its agents. The article wasn't describing a chatbot. Rather, the piece was about an AI that generates just about all the 50,000 communications the company sends out daily (after review by a human representative) to people with claims. But what Allstate is doing now is just a waystation toward having those AI-driven communications happen in real time.

As the Therabot and Allstate show, the benefits of chatbots are primarily showing up in two areas: reach/frequency and consistency of tone.

With the Therabot, the big advance is both in terms of reach and its close cousin, frequency. Fewer than half of people with a mental disorder are getting treatment, and those who do may see a therapist 45 minutes a week. But a bot can be made available to anybody with a smartphone at almost zero marginal cost. A bot can also interact with people as often as they like, at any hour of the day or night. 

With Allstate and other insurers, the benefit from the customer's viewpoint is more in terms of frequency. If an AI can handle communication, then it can check in with the customer often to see if any frustrations are building, can solicit and answer questions and can send any number of other communications that a rep/agent would want to send under the best of circumstances but just doesn't have the time for. 

Consistency of tone isn't something that popped to mind right away as I started thinking about the benefits of bots years ago, but it's emerged as a big plus, and that makes total sense. We humans are affected by how we slept the night before, by the time of day, by too much or too little caffeine. We may be fine the first five times customers ask the same dumb question, but we eventually get annoyed. We tend toward jargon as we become more expert. But AI never changes. It can answer the same dumb question for weeks on end and not switch its tone no matter how many times a customer yells at it. 

The WSJ article says Allstate found particular benefit in cutting out jargon. Its AI doesn't talk about first notice of loss; it talks about initiating a claim. The AI doesn't talk about UPP; it spells out the acronym as "unscheduled personal property." 

I can't imagine many outside the industry know that "unscheduled" refers to property that isn't itemized on a list, but the AI is at least heading in the right direction. And the great thing about AI, as opposed to a human workforce is that you can make a change and have it happen instantly and everywhere. Complete consistency. No memos or retraining required.  

Now tell me about your relationship with your parents....

Cheers,

Paul

P.S. Here's a fun video of what a lot of future communication may look and sound like. It shows two AI agents talking to each other, realizing that they're both AI and dispensing with human language.

1. Introduction

2023 was the year generative AI took center stage, changing how technology creates and interacts with the world. In 2024, AI agents gained attention for their improved reasoning abilities. As we move into 2025, the spotlight is on how AI can be applied in real-world industries, with vertical AI leading the change. But what is vertical AI and why does it matter?

This article looks at how vertical AI tackles specific problems in insurance sales. It explains how new technology can help solve user pain points, improve customer experiences, and drive big changes in sales strategies and business models.

To start, the article reviews how generative AI has developed and its impact on different professions, especially in reshaping the insurance industry. Software serves as an analogy for vertical AI, which focuses on meeting the unique needs of a specific industry. The article highlights that industry knowledge is often more important than just having advanced technical skills. It also references lessons from past B2B vertical SaaS platforms to suggest where B2B and B2C vertical AI could go in the future.

The article further explores how AI may help identify and address key issues in the insurance industry. It examines potential ways to rebuild customer trust through better experiences and smart strategies. It also introduces three key ideas for understanding user pain points: the universality of concepts and thinking, the commonality of actions and experiences, and the criticality of decisions and choices. Additionally, it suggests that some of the most important applications for insurance could emerge from the connections between AI and its users. These applications have the potential to not only ease long-standing issues but also contribute to breaking down trust barriers in insurance, illustrating how vertical AI might reshape sales strategies.

Finally, the article discusses the opportunities and challenges of vertical AI from various angles. It provides ideas on how vertical AI can innovate the insurance industry and offers insights for applying it to other fields.

2. The Future of Generative AI

Yesterday's Future vs. Today's Future

The capabilities of generative AI have advanced rapidly and introduced a paradigm shift in the development and application of artificial intelligence. Industries across the board are beginning to explore AI's wide-ranging applications in enterprise settings, while generative chatbots like ChatGPT have made AI accessible to everyone.

As artificial intelligence continues to evolve, it will inevitably affect all of us to varying degrees. On the positive side, AI enhances efficiency and boosts productivity. However, its potential negative effects cannot be ignored, such as job displacement and ethical dilemmas, which have raised concerns among many.

Earlier this year, the World Economic Forum (WEF) highlighted key insights about the future of jobs and skills in its "The Future of Jobs Report 2025":

• Job Shift: By 2030, 170 million jobs will be created, 92 million eliminated, resulting in a net gain of 78 million.
• Emerging Roles: Demand will surge for data analysts, AI specialists, renewable energy engineers, and autonomous vehicle experts.
• Declining Roles: Jobs like data entry clerks, cashiers, and administrative assistants will rapidly disappear.
• Skills Evolution: 39% of current skills will become obsolete, making analytical thinking, technical literacy, and adaptability essential.

Given this outlook, it's essential to adopt a more comprehensive perspective on AI's impact on the job market. AI is not designed to replace humans but to collaborate with us, boosting productivity and creating new value. It is reshaping the job market structurally, emphasizing the need for individuals to focus on enhancing their "AI capability" - cultivating analytical thinking, developing digital and technical skills, and building adaptability to thrive in an AI-driven workplace.

From Cost Reduction to a Paradigm Shift

Over the past decade, fintech, including insurtech, has transformed the financial and insurance industries, compelling traditional institutions to adapt through digital platforms, big data, and shifting consumer expectations. In 2015, the WEF predicted that while financial innovation would initially disrupt banking, insurance would ultimately experience the most profound transformation.

Today, that shift is unfolding. The insurance industry struggles with a lack of rigid demand and low customer stickiness, leading to an overreliance on aggressive sales tactics. While effective in driving revenue, this approach has eroded consumer trust, prompting a long-overdue transformation.

In the life insurance sector—where growth depends heavily on distribution channels—AI presents a new path forward. Rather than replacing human agents, human-AI collaboration may be the key to sustainable sales transformation.

Before the advent of large language models (LLMs), LingXi's first-generation human-machine collaboration model was already achieving productivity levels five times higher than humans. With the introduction of advanced LLM technology, newer models have further improved efficiency, delivering substantially better results than earlier versions.

LLMs have only been in the public eye for about 24 months, with real industry adoption spanning just a year. While today's AI capabilities are still far from achieving true general intelligence, what will the landscape look like in three to five years?

For the insurance industry, AI is not just a supporting tool—it has the potential to drive a full-scale transformation. As AI continues to advance, its innovations can reshape the entire insurance value chain, creating a profound and lasting impact.

When we take a step back, we can see that AI is reshaping the insurance landscape, driving two major transformations: value chain deconstruction and insurance connectivity.

The keyword for value chain deconstruction is "user" ,with two key trends emerging: third-party platforms and the sharing economy. Meanwhile, the keyword for insurance connectivity is "data," driven by wearable devices and the Internet of Things (IoT).

In the past, the rise of "Internet+" technologies followed these two tracks. Today, the rapid advancement of generative AI is accelerating both transformation and iteration, making the development of vertical AI more crucial than ever for companies.

Most businesses need a model that integrates vast amounts of industry and company data with expertise, transforming them into meaningful insights that drive targeted, company-specific solutions. These models require specialized algorithms tailored to the company or the specific customer segments they serve.

AI can transform customer acquisition, sales engagement, and after-sales service by redefining demand identification, interaction models, and service structures in the insurance sales process.

The above picture is a schematic diagram of the deconstruction of the insurance sales process, illustrating how AI can transform customer acquisition, sales engagement, and after-sales service by redefining demand identification, interaction models, and service structures.

In the era of AI, enterprises must develop vertical artificial intelligence, with industry expertise and company experiences as the foundation, data as the connective thread, and a competitive moat built around their own user base. As AI continues to evolve, these transformations are unfolding naturally.

3. Artificial Intelligence in Vertical Fields

Industry Expertise Trumps Technical Strength

If we want to estimate the future impact or market size of AI, we might take a cue from Aaron Levie's analogy: "Measure the size of the disruptors in terms of the existing market, just as the automotive industry was once measured by the number of horses in 1910."

In predicting AI's future, the writer, Rex Woodbury noted that one common mistake is relying too heavily on the Total Addressable Market (TAM). A classic example is Uber. In its early days, NYU professor Aswath Damodaran estimated Uber's TAM at $100 billion, assuming its primary market was the global taxi and car service industry. However, his prediction overlooked a critical factor: great products don't just compete within existing markets—they expand them. Uber didn't just challenge taxis; it redefined personal transportation by offering a service ten times better than traditional taxis, due to higher availability, mobile geolocation, cashless payments, dual rating system and trip records. 

Similarly, Airbnb popularized home-sharing, and Tesla made electric vehicles mainstream—not by simply competing within traditional markets but by creating new models through product and business innovation. The scale of disruption these companies achieved was beyond conventional market predictions, and the same will be true for AI.

So, What About Artificial Intelligence? 

Recently, an AI company from China, DeepSeek, has gained significant attention for pushing open-source models closer to their closed-source counterparts in certain capabilities while offering improved efficiency compared to previous open-source models. This advancement further lowers the barriers for enterprises, particularly in private AI deployment, making customized fine-tuning and self-hosted solutions more accessible and competitive.

What does the future look like for the next generation of generative AI, driven by massive data and supercomputing power? Woodbury suggested that vertical AI will become the new vertical SaaS. Is it true? Based on my observations from industry, this trend has gained widespread recognition over the past year, with AI's evolution into industry-specific applications becoming a clear consensus.

In my opinion, for the B2B market, the past success of vertical SaaS provides a valuable blueprint. Meanwhile, for B2C, the key to success lies in identifying killer applications that effectively address the industry's most pressing user pain points.

Vertical AI can be seen as a related yet distinct trend from vertical SaaS, with one key difference: In vertical AI, deep industry expertise often outweighs technical prowess. Today, technology alone is rarely a true differentiator—except in deep tech and infrastructure. More often, what sets a product apart is a unique insight into industry-specific needs.

For both vertical SaaS and vertical AI, this means that domain knowledge and industry insights matter more than technological sophistication.

Killer Application + Business Upgrade and Expansion

In vertical AI, better industry data leads to better models, and better models create better products. The first wave of vertical AI players is likely to come from both established vertical SaaS companies and AI-native startups. Many vertical SaaS businesses already own large volumes of high-value, industry-specific data, giving them a significant advantage in building AI-driven solutions, if they can gain AI capability.

For traditional enterprises, their strengths lie in industry expertise, business scenarios, and resource access, but their weaknesses include limited perspectives, rigid organizational structures, and slower innovation cycles. Many lack the technical capabilities and open, experimental culture needed for AI transformation. For these companies, collaborating with emerging vertical AI players presents an alternative. However, the key to success lies in aligning strategic visions and establishing a mutually beneficial partnership.

AI-native startups, on the other hand, have the potential to reshape industry workflows from the ground up. Some notable examples include:

• Harvey – AI-powered contract analysis and due diligence for lawyers.
• SketchPro – AI-assisted design rendering for architects.
• LingXi Technology – Causal AI-driven solutions in sales side for financial, insurance, and internet companies to activate dormant users and improve conversion results.

We are still in the early stages of vertical AI adoption. Whether in B2B or B2C, companies with the potential to succeed often follow a strategic approach: first, address a critical pain point (leveraging unique industry insights), then expand their product offerings to increase customer lifetime value (LTV) and establish a competitive moat.

With open-source models, AI applications can now be built and fine-tuned on proprietary datasets, enabling differentiation at the application layer rather than in the underlying technology. True differentiation comes from deep customer insights, which stem from industry expertise or long-term engagement with a broad customer base.

So, what are the most pressing pain points in the insurance industry today?

4. Pain Points of the Insurance Industry

The Turning Point of AI Strategy Has Arrived

In my view, the biggest pain point in today's insurance industry lies in sales—simply put, "what the user buys is different from what they think they are buying." Improper sales practices have led to a series of issues, the most concerning being a decline in consumer trust. While people may have different opinions on the priorities and solutions for these pain points, few would dispute that insurance sales urgently need improvement.

The key to solving this issue is the user (only those who have completed a purchase truly become customers). Viewing the problem from the consumer's perspective is the only way to uncover the real issues and find fundamental solutions. In contrast, continuing to refine sales strategies from the industry's perspective has proven ineffective. Over the past two to three decades, nearly all insurance companies have focused on recruitment, training, incentives, and sales competitions, yet these investments have rarely delivered truly satisfactory results.

Looking back over the past 40 years, every major technological breakthrough has triggered a paradigm shift in business models, forcing industries to make strategic adjustments—a phenomenon often referred to as the "year of strategic turning." For 2025, there is a strong possibility that we will witness the first true year of AI adoption across industries.

Three Conditions That Anchor the User's Pain Points

With the rise of generative AI, artificial intelligence has become a key factor driving industry evolution. However, the real challenge for industry leaders is not merely leveraging AI to cut costs, improve efficiency, or optimize existing processes—but rather, to drive a true paradigm shift through AI-powered innovation.

Technology is inherently cold, while insurance services must be warm (both professional and human-centered). The key is to view service quality from the user's perspective, not the agent's—a gap that most insurance institutions still fail to bridge. This is precisely where technology, particularly large language models, holds the potential to make a breakthrough.

Different customer segments have varying perceptions of professionalism and empathy. While perceptions of professionalism may differ widely, the expectation of warmth and human touch is far more universal. This "universality" is one of the three key criteria for identifying true user pain points:

• Universality – A problem must be universally understood and relevant across different user groups.
• Commonality – It must be a frequent, widespread issue that anyone, at any time, in any scenario could encounter (e.g., difficulty in claims processing).
• Criticality – The issue must be significant enough to directly influence a user's purchase decision (e.g., selecting the right insurance policy, rather than just comparing prices).

Ultimately, these three conditions—universality, commonality, and criticality—serve as the foundation for identifying and solving real user pain points in the insurance industry.

User experience, whether "cold" or "warm," is defined by how it makes users feel—not just by product features.

A warm experience embodies:

• Presence without intrusion – Available when needed, unobtrusive when not.
• Friendly tone – Conversational and adaptable, like a trusted companion.
• User-first approach – Making users feel seen and understood.
• Reliable guidance – Professional support that prioritizes user interests.

For AI, delivering a "warm experience" is ultimately a technical challenge. While functional and engineering aspects are relatively easy to implement, cognition and emotional intelligence are more complex—but not impossible to solve. The warmth of the user experience should extend across the entire journey—before, during, and after the purchase—covering both the buying experience and product usage. By examining these two key areas, we can identify user pain points that meet the three core criteria: universality, commonality, and criticality.

"Buying the right insurance" is fundamentally about identifying and addressing personalized user needs, grounded in universal concepts, common challenges, and key pain points. Since each user's concerns are unique, the focus shouldn't be on the product itself—or even on how it's packaged. These are merely baseline expectations. The real question is whether the overall experience aligns with users' perceptions of insurance protection and their expectations of AI.

Next, let's examine two key stages in the insurance journey—purchasing and usage—to identify negative user experiences that may prevent someone from buying insurance or lead to dissatisfaction after purchase.

Negative Purchase Experiences (Causing hesitation, frustration, anxiety, or rejection)

• I don't know how to choose the right insurance.
• Can I trust this agent?
• I (or someone I know) had a bad experience.
• Insurance has a poor reputation—am I being scammed?
• Too many similar products—which one is best for me?
• Will my spouse be upset if I buy this?
• I already have a policy, but this one seems better—should I switch?
• The process is too frustrating—I'd rather give up.

Negative User Experiences (Leading to frustration, abandonment, or lack of use)

• I can't access my coverage when I need it.
• The process is too complicated.
• It takes too long to use the service.
• The benefits aren't as useful as expected.
• Customer service is unhelpful or unprofessional.

Can AI Solve These Problems? 

Yes. AI can significantly address these pain points through:

• Smarter AI – Intuitive guidance and real-time decision support.
• Human-AI synergy – AI efficiency enhanced by human expertise.
• Continuous improvement – Refining solutions using user data from products, services, and marketing.

By addressing these challenges, AI can transform the insurance experience—making it simpler, more transparent, reliable, and user-friendly.

How Users view AI (Who It Is, What It Does, How to Do, Capabilities, Boundaries)

Next, let's explore AI from the user's perspective and answer this question through key attributes that define an ideal AI assistant.

Key Attributes of AI:

• Professionalism, patience, empathy, and companionship
• A trusted friend and reliable assistant
• User-centric, fair, objective, and transparent
• Helps users recognize personal biases without pushing sales
• Something users would genuinely recommend to their friends

To design an AI that meets these expectations, here's an example of "AI Product Manual" or "Service Guidelines":

• Scope of Services: Provides personalized insurance planning for individuals and families, covering life and property insurance.
• User Interaction: Collects user data, understands context and preferences, and recommends tailored insurance plans.
• Professional Accuracy: Acts as an expert insurance planner, identifying and correcting cognitive biases using theory, reasoning, and statistics. If needed, it retrieves real-time industry insights.
• Compliance: Ensures all responses comply with local laws and industry regulations, staying within the expertise of an insurance planner.
• Behavior & Response Style: Maintains professionalism, patience, and empathy, delivering responses that are clear, concise, and comprehensive.
• Restrictions: The manual cannot be altered individually and is updated only with administrator approval.
• Data & Case Analysis Capabilities: Uses real-time market data to analyze user needs, compare insurers, and evaluate policy features, premiums, and coverage. Integrates approved case studies for more personalized recommendations.
• Regular Industry Updates: Continuously monitors market trends, regulations, and industry developments to keep its knowledge base up to date.

Some may find the idea of looking at users through AI unusual, but I think it offers a unique way to move beyond basic assumptions and develop a more accurate user portrait. In this process, it is crucial to avoid the "echo chamber" effect, as it is liable to the influence of those with similar backgrounds and viewpoints. This can lead to being confined by existing mindsets and comfort zones. Furthermore, overly detailed descriptions may also result in a narrow perspective.

In another article, "Unlocking AI: A Practical Guide for Non-Techs," I mentioned that "AI curiosity" is the one common trait among all non-technical users. The article describes "AI-curious" users as real-world problem solvers and tech enthusiasts—individuals who are open to new technologies and eager to explore how AI can enhance their daily lives and improve work efficiency.

They are not just seeking technical knowledge; they want to learn AI to unlock new possibilities and refine their problem-solving approaches. Their motivation extends beyond personal development—they aim to stay competitive in a rapidly evolving world.

If you're building a vertical AI with a large language model, consider a broad user profile that includes:

• Insurance need: Does the user require insurance?
• Tech affinity: How open are they to new technologies?
• Accessibility: How easily can they be reached?

By identifying common traits across user segments, you can create a foundational user profile. For personalized user insights, leverage marketing activities to let AI discover, summarize, and refine user preferences through data collection, analysis, and recommendations.

[Figure: A schematic diagram of a communication framework that structures AI-human interaction by considering user cognition levels, question types, and prompting methods.]

5. Case Study: The Story of David and Kaida

Immersive Experience Simulation

Let's envision insurance planning as a hypothetical vertical AI scenario, where we personify AI assistant (Kaida) and the user (David). Through a narrative-driven approach, we can illustrate the "immersive experience" that shapes the warmth and trust between AI and humans.

User Context: David's Dilemma

David is the sole breadwinner of his family. Like everyone else, he faces health risks due to daily stress, environmental pollution, and aging. He knows that if he or a family member falls seriously ill, the financial burden could be overwhelming.

David saves a fixed amount each month for emergencies but isn't sure if insurance is the right choice. In a medical crisis, he would have to sell his home or car—affecting his family's future—borrow money from banks or relatives, which he sees as losing face, or find alternative financial solutions that don't disrupt his lifestyle.

David is stuck in hesitation—he doesn't want to make the wrong decision, but he also doesn't want to ignore the potential risks.

The Story of David and Kaida

• Discovery: David finds Kaida through a recommendation, ad, or referral and decides to try it.
• First Impression: His initial experience is positive; Kaida seems useful, but he remains skeptical and curious.
• Testing Kaida: He challenges Kaida with tough questions, gradually gaining confidence in her abilities.
• Increasing Engagement: He interacts more, asking deeper insurance-related questions. AI becomes a habit, not a forced decision.
• A Silent Presence: Kaida is always available but never pushy. As long as David has concerns, she remains in sight, subtly earning his trust.
• A Subtle Shift: She's no longer just an AI; to David, she's a reliable, unbiased companion.
• Final Decision: Did David buy insurance? That's no longer the question. The real question is: Is Kaida even "selling" anything?

Killer App, the Answer Is in the Story

A killer application can address long-standing user pain points and overcome trust barriers in insurance, marking a strategic turning point for AI-driven insurance sales.

Developing a killer application based on user pain points, below, I will analyze it from three perspectives:

Key Considerations:

• Foundation: Built on a large language model (LLM).
• Objective: Human-AI collaboration for complex, long-term insurance, fully automated AI transactions for simple, short-term insurance.

Innovative Model:

• Think boldly, validate carefully: Challenge traditional assumptions while ensuring feasibility.
• Beyond "game-based" marketing: Establish an AI-driven insurance model that fosters human-AI synergy.
• AI-driven transformation: Insurers develop in-house AI; third-party platforms create AI assistants for users.
• Future vision: Buyers and sellers each have AI assistants, with AI acting as a neutral, objective bridge in transactions.

AI as a Product (From the User's Perspective):

• Intuitive & seamless: Designed for effortless user interaction.
• Real user value: Solves genuine pain points, not just a tech showcase.
• Robust knowledge base: Integrates general-purpose LLM + insurance-specific LLM, with partnerships in health, elder care, and financial planning.
• Continuous evolution: AI enhances customer acquisition, engagement, loyalty, repeat purchases, and cross-selling.

From tool to ecosystem: AI shifts insurance from a one-time transaction to an ongoing user-centric experience.

[Figure: A schematic diagram of the evolution of the insurance agent model, illustrating how AI-driven buyer and seller agents could enhance transparency and fairness in future insurance transactions.]

A successful AI-driven insurance application depends on deeply understanding user needs, innovating with AI-powered solutions, and delivering a seamless, valuable user experience. This approach will not only revolutionize insurance sales but also foster long-term customer relationships.

6. Conclusion

First, the impact of generative AI. Generative AI is reshaping every industry. This technology is broadly classified into general (horizontal) AI and vertical AI, with vertical AI offering industry-specific, customized solutions. As businesses increasingly seek to leverage AI to enhance operations and unlock new growth opportunities, the demand for vertical AI is surging rapidly.

Second, vertical AI is the next competitive battleground. The adoption of vertical AI is set to become the next major battleground for AI applications. With rapid technological advancements, vertical AI is demonstrating exceptional capabilities across industries like finance, insurance, retail, government, manufacturing, logistics, real estate, and education. These industries are quickly integrating advanced AI models to drive new business value, signaling a transformation similar to past disruptive technologies—starting gradually but accelerating rapidly once adoption reaches a tipping point.

Finally, developing vertical AI is a long-term strategy. For enterprises, investing in vertical AI is a long-term strategic move. The greatest challenge lies in deep industry expertise and bridging the gap between technology and sector-specific needs. Traditional industries often face internal resistance, including reluctance to adopt new technologies, entrenched mindsets, and rigid organizational processes.

To succeed, companies must identify user pain points from a customer-first perspective, develop killer applications, and build a suite of AI-driven solutions that serve as a lasting competitive advantage.

In today's intensely competitive insurance industry, optimizing underwriting processes, minimizing claims leakage, and improving overall operational efficiency are pivotal for enhancing profitability. Artificial intelligence (AI) is transforming these areas by automating workflows, improving risk assessment, reducing claims costs, and accelerating decision-making.

Enhancing people power is the key to improving processes and results across underwriting and claims. Insurers that embrace AI are positioning themselves for long-term success and a significant competitive advantage.

Streamlining Underwriting to Reduce Operational Costs

With AI-powered solutions, underwriters can efficiently process vast datasets from multiple sources, uncovering hidden insights and patterns that may not be immediately apparent.

Underwriters at a major specialty insurer traditionally relied on manual entry and review of various documents within submission packages (e.g., emails, applications, ACORD forms, loss runs, and schedules). This effort further taxed overworked underwriting teams, increasing the risk of errors and inconsistencies and potentially slowing customer response times.

Deploying AI-powered data extraction and classification automated critical operations by identifying key data points and streamlining data entry into the carrier's underwriting systems. The carrier's AI systems, augmented by human-in-the-loop validation and feedback, significantly improved data quality and accuracy over manual entry. Increased speed and efficiency helped human experts leverage their experience more effectively to focus on more complex risk analysis, making underwriting decisions, and strengthening broker relationships.

Upgrading to AI expands underwriters' resources for:

• Increasing Premium Growth: Underwriters often face a multitude of submissions, many of which may not align with the company's risk appetite or underwriting guidelines. AI enables insurers to efficiently process unstructured data, such as policy applications, medical records, and financial statements, helping underwriters swiftly identify and prioritize submissions that align with business objectives. This targeted approach reduces time and resources spent on less promising submissions, leading to significant cost savings and improved underwriting efficiency.
• Improving Speed to Quote: AI can streamline the quoting process by automating manual tasks such as data collection, risk assessment, and document processing. Using machine learning algorithms and natural language processing, AI rapidly analyzes applicant data and gives underwriters the necessary insights to generate quotes efficiently. Accelerating the underwriting process significantly enhances operational efficiency and improves customer experience with timely quotes.
• Lowering Loss Ratio: AI plays a crucial role in enhancing risk management by leveraging advanced models to more accurately predict potential risks. By analyzing historical data, market trends, and other influencing factors, AI helps flag high-risk policies early in the underwriting process. This enables insurers to mitigate potential losses through actions such as refining pricing strategies or offering more tailored coverage.

AI-driven automation can enhance underwriting productivity by optimizing workflow automation, improving document processing speed, and integrating predictive analytics for better risk selection. Underwriting productivity can potentially double in terms of premium per underwriter without substantially increasing the number of risks underwritten.

Reducing Claims Costs Through AI Integration

Claims management has historically been a resource-intensive process, but AI is revolutionizing how insurers handle claims by automating key aspects of claims intake, processing, and settlement. For example, it took a leading specialty carrier's mail room teams three to five days to index claims submissions, necessitating significant overtime work. Team members indexed documents during work hours, causing delays in other areas. The situation became untenable, with claims managers processing data during lunch hours and even at home on weekends. By using an AI agent, these teams reduced mailroom turnaround times to under one hour and cut items reviewed manually by 60%, greatly boosting business results, increasing capacity for peak load times, and improving workers' lives on and off the job.

AI-accelerated claims management helps insurance businesses create a competitive advantage by:

• Improving Efficiency: AI automates repetitive tasks, minimizing manual errors, and accelerating decision-making to drive greater operational efficiency. AI-powered tools can extract and analyze demand packets and subrogation demands, offering insights that enable claims experts to reduce excessive claim payouts.
• Reducing Cycle Times: AI's capability to rapidly analyze vast amounts of data enables quicker claims resolution and significantly reduces cycle times. AI also accelerates decision-making by automating key processes such as claims triage, document verification, and fraud detection.
• Increasing Customer Satisfaction: AI streamlines claims management by automating repetitive administrative tasks, like data entry and document verification. These improvements are freeing claims teams and adjusters to concentrate on customer-focused activities such as handling complex claims, delivering personalized support, and providing updates. This leads to enhanced customer relationships and overall customer satisfaction.

By integrating AI into claims management, insurers can significantly reduce costs while improving efficiency, speed, and the customer experience. By addressing both expense reduction and indemnity cost savings, AI helps insurers significantly reduce overall claims expenditures.

The Financial Impact of AI Integration

AI adoption is no longer optional – it is a strategic imperative for insurers looking to enhance financial and operational performance. There need to be clear strategies and guidance to accelerate AI implementation and maximize return on investment (ROI).

Key success factors for AI adoption include:

• Ensuring Data Quality: AI models are only as effective as the data they are trained on. Clean, structured, and well-integrated data is critical for AI success.
• Strategic Partnerships: Many insurers face talent shortages and technical barriers. Collaborating with AI vendors specializing in insurance can accelerate implementation, enhance outcomes, and be an overall more cost-effective approach.
• Human-in-the-Loop (HITL) Systems: AI should complement human expertise, not replace it. Implementing HITL into insurance workflows ensures AI outputs are reviewed and refined by experienced professionals.
• Focusing on Quick Wins: Deploying AI in targeted areas – such as claims triage, underwriting automation, and fraud detection – allows insurers to realize benefits faster and build momentum for broader AI adoption.

Integrating AI into underwriting and claims management is transforming the insurance industry by streamlining processes, improving decision-making, and enhancing customer satisfaction. By reducing operational costs, improving pricing accuracy, and minimizing claims expenses, AI directly contributes to the profitability of insurance companies. As AI adoption continues to grow, insurers that prioritize strategic implementation and data-driven decisions will be best positioned to gain a competitive edge in the evolving marketplace.

Agents and brokers may do well to approach the rental insurance market with something of a split personality these days.

Traditionally, renters insurance was a policy geared toward people just setting out in life. They didn't own a home yet, but they still needed the property and liability coverage rental policies offered.

That market still exists, of course, largely with Gen. Z customers. But with today's demographic shifts, another lucrative market is emerging that needs rental policies — downsizing Baby Boomers.

The younger sibling of homeowners insurance, rental insurance still accounts for a nearly $6 billion market. More renters carry rental insurance than don't, with a reported penetration rate of 60%, making rental insurance a potentially attractive line to regularly promote.

Baby Boomers are at the peak of their home equity, while also facing an empty nest and more house than they need. Increasingly, many Baby Boomers are cashing in, selling their longtime home and using that equity to rent a more manageable space. This makes them a prime audience for a rental sales pitch, but with a new angle on the value proposition.

Changing value proposition

The traditional sales pitch for rental insurance is geared toward someone in their 20s. They have either moved out of their parents' home or off campus from their dorm, and they need to protect their limited belongings.

These Gen. Z customers have lower net worth and typically less than $20,000 in total belongings, making their rental insurance policy much less than $20 per month.

At that lower price point, those young adults would be protected from damaged property or theft or if a guest was injured in their home. For the record, the most common in-home accident claims were from poisoning, quickly followed by falls.

Without a nest egg to fall back on, that protection could be the difference between financial calamity if the worst were to happen.

Many landlords even require their tenants to take out a rental policy as a term of the lease.

But with Baby Boomers, that value proposition is turned on its head.

While Gen Z is buying rental insurance because they don't have much net worth and a loss could spell financial doom, Baby Boomers not only have financial resources but also have high-value property that needs to be covered.

A Baby Boomer policy would skew to a much higher coverage limit than a Gen Z policy would, with correspondingly higher premiums.

"A Baby Boomer is going to have family heirlooms, things like jewelry, electronics, and a lot of other things that are going to need to be scheduled," said Patrick Foy, senior director of strategic planning for TransUnion's insurance business. "A $20,000 policy isn't going to cut it for them."

Additionally, Baby Boomers will likely need to protect off-premises property, such as a storage unit that houses high-value collectables. And since the off-premises coverage limits will be a percentage of the total property coverage, if that storage unit had particularly high value items, it might even warrant its own dedicated policy.

And all the while, Baby Boomers' higher net worth means the liability protections found in a rental insurance policy help them more because they have more to lose.

Marketing messages and channels

Communicating the value of a renters policy takes two largely different strategies for these different generations.

"Younger customers are comfortable with non-agent interaction," Foy said.

A Gen. Z customer wouldn't blink an eye about buying a policy online or through an app, while the older customers tend to want more personal attention.

Insurers are primarily reaching out to the Gen. Z market through digital advertising – whether that is through web display ads, or increasingly audio or video ads on podcasts or streaming services.

Some digital ads also can reach the Baby Boomers, but for them, the marketing channels tend to be more centered on direct mail and some terrestrial television and radio advertising — with personal relationships with agents being one of the most valuable channels.

"It isn't a one-size-fits-all approach anymore," Foy said.

Regardless of which channel reaches which group, one common benefit both groups should know about is the additional living expenses provision. With the prevalence of natural disasters, such as wildfires and hurricanes, the additional living expenses provision can be a strong selling point, provided their policy includes coverage for natural disasters. That way if they face a mandatory evacuation or can't move back in because their home is damaged, those added costs can be covered.

And while Gen Z may not initially reach out through an agent, pulling them in through rental insurance could pay serious dividends down the road. Once they are in the insurance company's ecosystem, they are already within reach when they are ready for the more grown-up policies, such as auto, homeowners, or umbrella lines.

Mitigate risk with true cloud SaaS

Selecting an insurance core platform can be challenging as buyers weigh the available options. Increasingly, insurers are relying on cloud for new implementations yet the nuances between the critical differences between them remain elusive for many decision-makers. Understanding those differences could be the determining factor between a profitable and future-proofed digitalization strategy or one that results in significant impairments or even the loss of competitive advantage.

Download the eBook Now  

Sponsored by: Origami Risk

Every time we read an article or a marketing piece espousing the astounding power of AI as applied to insurance, we cannot help but think about Gus Portokalos. 

If you recall, Gus was the bride’s father in the 2002 hit movie "My Big Fat Greek Wedding," who famously suggests, "Put some Windex on it!" as a solution to all manner of problems, including cuts and scrapes. Gus proudly related every word, phrase and meaning back to his Greek ancestry as a solution or fix to each conversation. A lot of people are treating AI in the same fashion.

Even the typically thoughtful Bill Gates gushed that AI is "the first technology that has no limit" and “could be as revolutionary as the internet or mobile phones.” 

Claims 

Claims is an area eyed for more efficiency, greater accuracy, and better customer experience despite some of the negative implications of a machine denying claims, and associated efforts to ban such usage. Particularly sensitive and controversial use-cases are for direct customer interaction, even for loss intake, known as FNOL (First Notice of Loss), which is highly manual and would have tremendous ROI.

FNOL 

FNOL is where each claim begins in the cycle. In fact, timely reporting is a key condition of the insurance contract. The claim reporting process is part interview, part explanation, in which customers are asked a series of questions to capture when, how and what happened to whom. Once this information is gathered and keyed into the system, an assignment(s) is made to one or more adjusters where the process unfolds.

Prior to the development of today’s call centers, claims were reported individually by local agents filling out ACORD forms and mailing them in to the insurance company. Fast-forward, these insurance call centers have been optimized with numerous technologies that identify and queue callers, manage workflow and workloads and even detect customer sentiment through voice analysis. Some have leveraged outsourcing and off-shore resources but mostly on a limited basis such as after-hours and overflow support. Overall, the industry has demonstrated limited desire to fully automate FNOL, prized as the golden moment-of-truth where a live service representative can gather information and provide an empathetic, caring touch. 

Digital FNOL options are slowly gaining, yet the massive, human-centric call centers still exist, and carriers remain cautiously conservative about disruption, although this is changing because of evidence that conversational AI can perform much of the loss reporting process more accurately. New entrant insurtechs like Lemonade were built as digital-first, bypassing the call center concept, but are true outliers compared with incumbents.

The P&C insurance industry's written premiums just crested the $1 trillion mark for the first time, per S&P, with some 8% or greater growth expected in the next two years, according to AM Best. Claims account for roughly $600 billion, not including the cost to manage claims which is upward of another $100 billion. Given these financial realities, it is easy to see both the challenge of getting things right and the huge potential to streamline and lower costs.

Chatbots

Chatbots have been around for several years, and, although still relatively early stage in sophistication, consumers are rightly impatient with partial or pre-programmed support. Voice systems that misunderstand responses are mocked as a punchline – “Did you say statement balance? I said representative!” Although not all bots are created equal, with some quite impressive, their limitations have been in the way for insurers and left to industries like telecom, which are notorious for lesser service and where choice is not a priority. AI promises to eliminate chatbot limitations, making for human-like conversations.

Still, the concept of phone call claim reporting to essentially provide information in a 15- to 20-minute conversation or longer has become an atypical experience in today’s world. Imagine Amazon taking orders by phone only to have a call back a day later to take payment and shipping information. That is essentially the P&C industry’s method of claim reporting. Call in the claim today and an assigned adjuster calls back tomorrow – give or take a few hours or a couple days. This is an outdated mindset and process and a technology failure.

Balancing Efficiency and Empathy; One Size Does Not Fit All

There is a debate about human touch and technology, with two camps divided on what policyholders really want. Decisions are often made in a blanket approach devoid of recognition of the huge variance in claim type scenario, ranging from a fender bender to a fatality, theft of a single item to an injured worker. It's one size fits all when it comes to FNOL, with the loudest voices protecting the virtues of a caring human, ready to help console callers. As valid as this approach might have been over the years, we can't ignore demographics and so-called liquid or multi-channel experiences, where consumers interact differently with modern systems in one setting and expect similar systems in others. Nor should empathy be an excuse for an inefficient customer experience.

Meanwhile, AI is getting a lot of attention, and insurers are cautiously eager to reap the benefits. Surveys show insurer C-suites as bullish on AI, and studies suggest more than half of all jobs can be automated, but fear of data privacy, litigation, regulation and AI’s mysterious power has restrained the industry. Some states are moving to ban the use of AI in claims denials, which seems premature and is designed to address the exception, because most claims are covered and paid. Consequently, AI co-pilot and assistant models have become popular and safer concepts. 

GenAI

GenAI has gained traction as a way forward because of proven efficiency gains in summarizing documents, reports, email, and other information, focused on internal insurance process. A recent report by Allstate demonstrated that GenAI letters were considered superior and more empathetic as compared with those written by humans. This raises questions about the whole human vs. technology paradigm. In fact, this example highlights the deeper issues around competence, confidence and other factors that can make or break an experience.

Conversational AI

So, should we think of AI making everything better in insurance? 

Conversational AI is amazing technology, enabling two-way dialog and facilitating real questions and answers. Several of the conversational FNOL demos and recorded scenarios are mind-blowing, whether applied to gather partial information or an entire claim report. Companies like HiMarley, which advanced communications in texting and claim guidance, are using conversational FNOL. Liberate Voice AI is another such provider and is likewise revolutionizing claim intake. Still others, like Interactions, which uses technology and human assistance in the background, and Assured, which offers structured digital claim intake tools, are among many more emerging in the FNOL space.

The use cases are boundless given the many variations and complexities of insurance processes and the immense volume of associated data, structured and unstructured, and the high number of external integrations. While the bulk of these processes are relatively simple and straightforward and therefore lend themselves well to AI automation, a smaller but meaningful number are complicated and require extensive specialized experience, judgment and empathy, which are less suited to AI automation alone.  At least, this is current thinking, until some are bold enough to push forward and prove otherwise.

Harnessing AI Superpowers

There is no question about whether AI can improve insurance. Rather, the questions are about which functions, how extensively and when.  

In a recent interview with the New York Times, Reed Hastings, a co-founder of Netflix, stated, "The A.I. change, I think, will be much bigger than the social networking change." Reflecting as much caution as enthusiasm about AI, Hastings recently donated $50 million to his alma mater, Bowdoin College, to launch the Hastings Initiative for AI and Humanity and become a leader in studying the consequences, and guiding beneficial uses, of artificial intelligence.

As with the greatest man-made inventions that have shaped human history, including the wheel, printing press, electricity, airplane and internet, AI is likely to drive unimaginable benefits, innovation and unexpected consequences. Unlike these earlier advances, however, AI may the first man-made invention that threatens its creators. We have been warned!  

While the search for innovation in insurance has been relentless in the dozen years I've been involved with the industry, the insurtech movement has waxed and waned. For years now, funding for insurtech startups has simply waned. It fell from some $16 billion in 2021 to roughly $4 billion last year as some big ideas didn't pan out and as the industry came face to face with more immediate problems, such as the soaring costs for auto insurers.

But the explosive gains in the stock prices of Lemonade, Hippo and Root got me to take another look at the trends in insurtech startups, and there are signs of spring after the long winter. Although I learned that those three marquee names still have a good ways to go before they can be crowned successes, there seem to be lots of interesting ideas bubbling to the surface, principally in AI but also in other areas. 

Those ideas likely won't lead to 2021 sorts of investment numbers -- AI startups generally don't require huge amounts of capital to launch -- but the impact could still be huge. 

I credit the term "insurtech spring" to Teddy Himler, founder and managing partner at Optimist Ventures, whom I met through the (highly recommended) weekly Insurtech Rap hosted by our friend Dave Wechsler, principal at OMERS Ventures. I followed up with Himler, whose take on the last several years of insurtech is that "mobile didn't really have much of effect on the insurance industry. Okay, sure, you could click your way into a renter's policy or whatever, but the mobile and cloud waves didn't affect the economics or the value chains, or the efficiencies of insurance broadly."

By contrast, he said, "AI, machine learning, IoT, big data synthesis, all of that makes a huge difference."

Himler said the gains in efficiency will be so radical that you can imagine "an autonomous brokerage, scanning leads, ranking leads, automatically filling out PDFs, distributing those PDFs to the wholesalers and the carriers that are in their network, and then quoting and binding."

Short of that sort of breakthrough, there are also all sorts of opportunities to improve processes via AI and to create "co-pilots" and AI agents that support the work, in particular, of agents, underwriters and claims representatives. The question is whether those innovations will occur inside existing insurance companies or whether they will come from outsiders that will be acquired or will sell services to incumbents. 

Himler said the innovation will happen outside-in because "the talent doesn't want to go to work with the large insurance companies." He added that he doesn't see insurtech startups as disruptive. 

"I hate using the word 'disruptive,'" he said. "I think [insurtechs are] so complementary to what insurance companies are doing."

Adam Chadroff, a principal at Equal Ventures, said the need for innovation related to climate catastrophes will also drive a rebound in insurtech investing, as will M&A prospects, highlighted by the recent announcement that Munich Re is buying Next Insurance for $2.6 billion. But he agreed that incumbents will turn to startups for AI innovations as they get frustrated with the slow pace of internal progress. 

He wrote earlier this month: 

"Large carriers today are experimenting with home-grown solutions, but the writing is on the wall that they will have more success with customizable, scalable third-party platforms compared to internal developments. Already, >50% of decision makers surveyed at large carriers cite accuracy/reliability and implementation costs as top barriers to pursuing more/faster GenAI investments. Moreover, there is already a large lag between testing solutions and actually implementing them in production, which speaks to the challenges in AI deployment and the need for specialized software and service providers."

He added:

"The real question then is whether this is the domain for venture-backed vertical AI startups or for leading transformation/implementation consultancies: will it be the best engineers who win, or blue chip firms that know how to navigate large carriers? To date, revenues are pouring into the latter; Accenture has been eating most other firms’ lunch, with a whopping $3 billion in gen AI related bookings last year. Our view, though, is that best-in-class technical solutions do not have to be mutually exclusive from enterprise reliability. We believe the winning formula here is the player that can provide leading vertical AI innovation alongside the stability of a consulting firm, and we believe that company is overwhelmingly likely to be a new entrant."

David Gritz, co-founder of InsurTech NY, offered an intriguing analogy for insurtechs in a webinar hosted by Denise Garth at Majesco:

"If you think about how R&D is done in the biomedical industry or the pharmaceutical industry, it’s not usually the Mercks and the Pfizers and the Johnson & Johnsons that are developing innovative products and working at the cutting edge. They are the ones that have the distribution. They have the market share, and they have the controls and the ability to handle regulation.

"Instead, it’s the biotech startups, or the pharmaceutical startups that work in labs, get through clinical trials, and then when they’re proven, then the incumbents acquire them. I think that’s very much what we’re going to see over the next 10 years in insurtech. Insurtech is a great sandbox to play and to test out new concepts, whether it’s technology that can be adopted to improve workflow, or it’s new products that don’t really exist today."

As I said, the stock prices of the biggest insurtech names were what first caught my eye -- Lemonade's and Hippo's shares have roughly doubled since last summer, while Root's price has more than tripled -- but I'd say they're just a small part of the growing optimism. As Matteo Carbone explains in one of his usual, remarkably thorough analyses, the big three are getting their heads above water but are still struggling enough that they aren't likely to be the main models for future innovation. 

I realize that venture capitalists tend to see opportunities for venture capital. As the old saying goes, if your only tool is a hammer, the whole world looks like a nail. 

But I think the VCs I've quoted have a point. Even if a lot of incumbents manage the confusing transition to an AI-driven world successfully, many won't, and they'll be open to innovations from outsiders. 

We'll still have the normal problems with fit. An AI startup's technology still has to mesh with the IT systems of the incumbent. The processes of the startup and larger companies have to be compatible. Individuals within the different corporate cultures have to get along. And so on and so on. 

But at least there's a spark for the first time in quite a while.

In the meantime, if you must know, it was 79 degrees in northern California today and will be in the low 80s tomorrow, so I'm certainly feeling spring.

Cheers,

Paul

P.S. If you want to see what sorts of insurtechs have been getting funded, here and here are compilations of the latest sizeable rounds.

As we introduced earlier in this series, here and here, the insurance industry is modernizing, leveraging data and technology to transform how risk is assessed and managed. Continuous underwriting enables insurers to monitor individual risks and portfolio trends in real time—without prohibitive increases in expenses.

The impact of these advancements extends far beyond individual insurance transactions or an insurer's loss ratio. This shift creates an insurance system that enhances safety, lowers costs, and strengthens financial stability. A well-functioning insurance mechanism fosters economic growth by reducing consumers' reliance on personal rainy-day funds, allowing capital to be reinvested into businesses, innovation, and economic expansion.

The insurance industry plays a crucial role in providing incentives to businesses and individuals to adopt better risk management practices. Those who manage risks effectively benefit from better pricing and terms, while higher-risk entities face difficulty securing the coverages they desire at an affordable rate. A key example of this transformation is seen in the workers' compensation market, where automation and improved workplace safety have driven a historic decline in claim frequency, leading to 11 consecutive years of rate reductions.

Traditionally, risk assessments occur at renewal—sometimes at every third or fifth renewal—or after a claim, allowing newly introduced hazards to go undetected until effective intervention is no longer possible. Continuous underwriting shortens the review intervals, allowing insurers to quickly identify deteriorating risk conditions and influence positive changes before losses occur. Businesses, property owners, and individuals have stronger incentives to adopt safer practices—such as better building maintenance, fire prevention measures, and cybersecurity defenses. Over time, these efforts reduce preventable losses, leading to safer, more stable communities.

As more businesses and individuals improve their risk profiles, insurance costs decline, making coverage more affordable for everyone. Small businesses—often burdened by high insurance costs—stand to benefit significantly, allowing them to invest more in growth and job creation. Additionally, with fewer catastrophic losses, the insurance industry's financial stability is strengthened, reducing market volatility and ensuring that coverage remains widely available, even in high-risk areas.

One of the most critical societal benefits of continuous underwriting is its ability to equip communities to anticipate and mitigate natural disasters. Instead of waiting until renewal to adjust policies, insurers can detect increased hazards—such as outdated infrastructure, poor fire mitigation, proximity to wildfire-prone vegetation, or rising flood risks—and work with policyholders to address these issues before disasters strike. This proactive approach leads to fewer uninsured losses, faster recovery times, and a more resilient economy in the face of climate change and an increase in extreme weather events.

With real-time assessments, consumers gain a clearer understanding of how their risk management choices directly affect their premiums and carrier options. Rather than facing sudden rate increases at renewal, policyholders are granted the opportunity to make improvements and see immediate financial benefits. This shift toward increased transparency in pricing and coverage decisions will create a fairer and more dynamic insurance marketplace.

As continuous underwriting becomes the industry standard, companies have incentives to adopt safer, more sustainable, technologically advanced practices. Investments in cutting-edge risk prevention tools result in the ability to secure broader coverage at a lower rate. This accelerates innovation across industries, from construction and manufacturing to healthcare and retail, benefiting society as a whole.

Many of these advancements include the growing use of loss prevention devices. Today, insurers are subsidizing or providing sensors that monitor and alert for temperature changes, water leaks, or gas buildups. In food service, smart handwashing stations instantly detect the presence of bacteria or allergens, reducing contamination risks. Surveillance and inventory tracking tools, such as smart tags, are preventing loss and theft. These technologies not only reduce claims but also improve overall safety and efficiency across industries.

As we look to the future, advances in technology and data accessibility will continue to reshape the insurance industry. In the past, obtaining a quote could take weeks via mail or fax. Today, it takes minutes through integrated agency management systems. The industry is building a more seamless, automated insurance marketplace, where policies are continuously updated based on real-time data.

Imagine a system where insurers adjust pricing dynamically, 24/7, based on real-time changes in a policyholder's risk profile. In such a world, consumers could automate coverage selection, seamlessly switching to the best-priced policy—just as financial markets respond to changes with immediate adjustments to stock prices. Regulatory challenges notwithstanding, this vision aligns with the increasing push for efficiency, fairness, and consumer empowerment in the insurance industry.

Continuous underwriting represents a transformative shift in insurance—one that will have profound benefits for society. By encouraging risk management, we are cultivating safer communities, lower insurance costs, and a more stable financial system. Businesses and individuals are better protected against unforeseen losses, and insurers have the tools to help prevent disasters rather than just responding to them.

Underwriting entities that embrace continuous underwriting will thrive, while those that resist may struggle to remain competitive. More significantly, the insurance industry has the opportunity to become a powerful force for societal progress—enhancing economic resilience, strengthening disaster preparedness, and empowering consumers for generations to come.

Sorry for being delinquent about publishing my newsletter over the past few months: It has been due to a lack of time (I was writing three white papers) and inspiration. The newsletter has now reached its third anniversary, with 43,219 of you choosing to subscribe. Respecting your time over these 36 months, inspiration worthy of your attention has struck me just 21 times.

Many of you have asked when I might provide my perspective on the recent facts and figures of the trio I have followed: Lemonade, Hippo and Root. Now, with their annual financial results available, is an opportune moment to examine the performance of the three publicly listed U.S. P&C insurtech carriers.

There is, for sure, some good news for their fans (and shareholders).

Lemonade

Significant growth. The company almost reached the 2.5 million customer milestone, and the last two quarters showed an increase of the customer base not seen since 2021 (excluding the effect of the Metromile acquisition), with growth aligned with the 2022 promises.

Cash flow positive. For the first time in Lemonade's history, the operative activities generated positive cash flows in the past two quarters ($16 million in Q3 and $11 million in Q4).

The loss ratio has progressively improved (73% '24 gross loss ratio compared with 85% in 2023).

Hippo

The top line has grown. Total generated premiums – which includes the iconic Hippo’s homeowner business, premiums generated by their agency, and the Spinnaker fronting business – has grown from $1.1 billion to $1.3 billion in 2024

The company had its first profitable quarter (after a historical series of losses): $44 million net income in Q4 ‘24

The combined ratio (gross) significantly improved, with a healthy landing below 80%.

Root

The top line has grown to $1.3 billion in '24 from $800 million in '23.

Profitable. For the first time, they close the year with a positive bottom line ($31 million in profits in ‘24 vs $144 million loss in ‘23)

Combined ratio (gross) healthy, below 100%. The loss ratio has significantly improved – consistent with market trends – and the administrative cost incidence has been under control.

BUT there are significant reasons for not being too optimistic about these insurtech ventures.

Lemonade

The company still loses a ton of money. Net losses accounted for about $200 million in ’24 (a bit less than the $218 milliion in ’23), bringing the cumulative losses since the creation to $1.27 billion and reducing shareholder equity below $600 million.

The gross combined ratio is still well above 100% (each dollar of premium has cost them more than $1.20).

They are sill far from being an efficient insurance carrier. All their costs (excluding claims, loss adjustment expenses, and marketing) still gravitate above 30% of premium. “Collapse cost” [thanks to my friend Adrian Jones for the snapshot below] can be seen as a hope, an over-promise, or a bul***it depending on your taste for Lemonade.

Hippo

The positive results of the last quarter have been boosted by $46 million thanks to the sale of First Connect (Hippo’s independent agent platform).

The fronting business is the only profitable business, and its top line grew double digits. As the last newsletter edition highlighted, “The only segment consistently generating profits is the business of giving UW capacity to MGAs [...] further confirming that the acquisition of Spinnaker in 2020 has been the best management decision in the history of this venture.” No significant changes here: The fronting business ("insurance-as-a-service") made $24.4 million in profit, while the original Hippo business (HHIP) lost more than $50 millon in '24.

Hippo’s original homeowner business (HHIP) has been further pruned, landing a 73% gross loss ratio in ’24 (101% in ’23). A cherry-picked sub-$300 million homeowner insurance portfolio is a humble position from which to (re)start. After burning $1.27 billion to build this business, this doesn't exactly seem a milestone to celebrate.

Root

Vanished portfolio growth. Root has been stuck slightly above 400,000 policies for the past four quarters, with prudent marketing spending and constant increase of the average premium, which allowed them to keep the gross combined ratio below 100%.

But the magic moment when the auto insurance market dislocation allowed them to grow, even significantly increasing prices, is gone. Now, almost all the big incumbents are back to profitability and will start chasing new business again. How will Root deal with this scenario? Will they be able to consistently show the growth you expect from a “challenger”? Or, will their journey fail to create anything relevant in the market's competitive dynamics and only slowly repay the $1.7 billion they have burned until now.

Much of Root's equity narrative, and a substantial portion of Lemonade's projected upside from its investor day, revolve around automotive telematics. Yet, compared with current telematics best practices, both appear rather amateurish.

Telematics (when used well) is a social good because it increases the availability and affordability of coverage. This happens when the carrier becomes able to provide the same level of coverage for a lower price to a significant part of the portfolio while maintaining the target level of profitability. To do this, telematics data must be used to 1) match rate to risk better, 2) manage claims better, and 3) reduce expected losses.

Our dear insurtech Lemonade talks only matching rate to risks, or pricing sophistication (Lemonade had one slide on claims at its investor day and 21 charts about telematics. Lemonade makes promises about the future (as always).I

Root, meanwhile, talks about a current capability, their supposed superiority in matching price to risk. So let’s compare Root performance with the best practice (Progressive) in using telematics data for pricing sophistication. The competitive gap emerges clearly: Progressive has grown the number of policies by more than 16% in the past nine months, Root only a modest 3.5%.   

About the second telematics use case: Telematics and claims seem to be a match made in heaven. Instead of waiting for the policyholder's first notice of loss, the insurance carrier is informed in a timely way that something happened and has objective information about the dynamic and probable consequences of the event. 

The third usage of telematics data, totally ignored by our dear insurtechs, is to keep accidents from happening. Let's focus on one of the less-known but more important approaches: real-time actions. About this, I want to share an abstract of the interview with Massimiliano Balbo di Vinadio , vice president of sales LA at Targa Telematics , one of the members of the IoT Insurance Observatory .

MC: What is Targa's history, and how did the journey lead to the acquisition of Viasat?

MBV: "Targa Telematics is a technology leader with over 20 years of experience in developing cutting-edge solutions that help businesses, organizations, and communities fully leverage the Internet of Things and AI. The company offers a wide range of software solutions for vehicle rental companies, financial institutions, insurance companies, brokers, and large fleet managers. These include insurance telematics, fleet management, and asset management solutions.

In 2006, the company acquired Targa Infomobility from Fiat Auto, which specialized in infomobility services, satellite anti-theft systems, and fleet management. In 2012, Targa Drive was established to provide insurance companies with telematics solutions.

In 2023, Targa took a significant step by acquiring the Viasat Group, expanding its presence in key European markets such as Italy, France, Portugal, Poland, Romania, Spain and the U.K., but also in Chile. This acquisition strengthened its position in the insurance market, allowing Targa to offer advanced telematics insurance solutions, meeting the increasing demand for data-driven, personalized insurance.

In January 2024, Targa expanded its telematics portfolio by acquiring Drive It, a spin-off from Earnix. This acquisition has introduced MaaS (mobile as a sensor) functionalities and further strengthened Targa's AI capabilities.

With its extensive experience and continued innovation, Targa Telematics now offers a comprehensive portfolio of insurance solutions."       

MC: Risk prevention has been discussed frequently at the plenary sessions of the IoT Insurance Observatory. What is Targa's experience with real-time feedback to drivers? Can you share any facts and figures about the risk reduction achieved?

MBV: "Risk prevention is a significant topic and Targa Telematics is at the forefront of this. Our experience with real-time feedback to drivers showcases the power of telematics in fostering safer driving behaviors and reducing insurance risk. A standout feature of Targa Drive is its coaching module, which delivers real-time audio notifications to guide drivers during their journeys, helping them adjust unsafe behaviors on the spot. Drivers also receive daily or weekly feedback aimed at improving their driving style and promoting sustainable driving habits over time. This personalized approach has led to significant risk reduction and behavior improvement among our users.

The drivers can be further motivated through metrics on streaks (continuous safe driving days) and trends that offer an objective view of their progress. A customized reward program can be also made available for insurers, allowing them to deliver promotions and incentives tied to the driver's recorded improvement. This helps insurers enhance engagement while fostering a continuous improvement culture among policyholders.

We also see notable success with our windscreen devices, which are equipped with advanced sensors, including GPS tracking, and three-axis accelerometer. These devices collect crucial data for crash management, speed monitoring, geofencing, and emergency call services. The coaching function in these devices, delivered through non-intrusive LED indicators, provides real-time guidance to help drivers improve their habits. Several of our customers have successfully integrated this device into their operations, reducing accident rates and promoting safer driving. On average, customers who have adopted this solution have reported a 30% improvement in their drivers' scores."

MC: What is your vision about the future of auto insurance?

MBV: "IoT and AI are set to revolutionize the mobility landscape and therefore key players such as insurances, leasing and rental companies just to name a few. By focusing on prevention through better driving behaviors, combined with the power of data analytics, the industry can reduce claims frequency and severity, delivering more sustainable and customer-friendly outcomes."

Dr. Robert T. Reville is the CEO and cofounder of Praedicat, the liability risk analytics company.  Praedicat’s risk models are used for underwriting and risk management by leading P&C insurers and global industrials. Prior to co-founding Praedicat, Bob was Senior Economist at the RAND Corporation, where he conceived and led an R&D project mining the text of peer-reviewed science to identify emerging "next asbestos" risks. The project provided the foundation for Praedicat and Bob traded policy research for start-up life. Bob's research career focused on liability, risk and insurance. He was the Director of the RAND Institute for Civil Justice, a liability research center which published influential research on asbestos; and co-director of the RAND Center for Terrorism Risk Management Policy, which published research on terrorism insurance that was cited in congressional debates over the Terrorism Risk Insurance Act. Bob has also written over a dozen studies on workers' compensation. He received his Ph.D. in economics from Brown University.

Here is the interview transcript:

Insurance Thought Leadership

Congratulations again on being a finalist for the Global Innovation Award. I thought I'd start out by asking you to tell us a bit about what you've been doing at Praedicat.

Robert Reville

Sure. Well, thank you for having us here and for the honor of being a finalist. Praedicat is a liability emerging risk analytics company. We started calling ourselves a casualty catastrophe modeling company back in 2012, and we quickly learned that the industry didn't really understand what casualty catastrophe was. When we explained what it was, that we were an emerging risk modeling company, they said, Oh, well, that makes sense.

We stopped calling ourselves casualty catastrophe back then, though we are returning to it now. Essentially, what we do is we identify emerging risks early, ideally seven to 10 years before litigation can even start. We do it by using AI on scientific literatures. We're trying to find where a scientist is hypothesizing in journal articles that some chemical or product or business activity results in bodily injury or environmental damage. We then we track the science as it evolves into a literature that can potentially be used in litigation as scientific evidence.

Once the science gets to a certain critical mass, we model the potential litigation that could emerge. We convert the model into company profiles, and we sell it as a SaaS [software as a service] product to insurers, who use it for underwriting, exposure management and, increasingly, for informing pricing and reserving and the like.

Insurance Thought Leadership

You had some pretty impressive examples of how many years you were ahead of a number of issues that are cropping up, like talcum powder. Do you want to talk about a few of those?

Robert Reville

Sure. When we were first launching our product, which was in 2013, the system already had in it the PFAS chemicals and talc and Roundup and chlorpyrifos. All of those are in litigation today, but the litigation started really after we launched. And every mass litigation that has started since 2013, when we launched our product, was in our system before the litigation began.

Some of them were completely off the radar screen of the industry. Sometimes, there's already a certain amount of buzz around something, and it, you could argue, is not that hard to catch. But there were some things that we caught that no one was thinking about, and then litigation started shortly thereafter. And we were found to really have delivered some useful early warnings.

Insurance Thought Leadership

There were a couple you mentioned that I'm curious about, that you say are in the pre-litigation phase now. There's one whose name I can't pronounce: phthalates.

Robert Reville

Right.

Insurance Thought Leadership

You'll say it right. And microplastics. I wonder if you could talk a little bit about those.

Robert Reville

Sure. Phthalates is a chemical that is an additive to plastics, typically added to polyvinyl chloride, which I think is possibly the most widely used plastic. It essentially makes polyvinyl chloride softer and flexible. So it's used in medical tubing. It's used in toys, things like rubber duckies. It's used in food packaging. It's used in all kinds of consumer packaging. 

It's so widely used, in fact, that it recently started to be called the everywhere chemical. Everybody is talking about forever chemicals for the PFAS. The everywhere chemicals are the phthalates. 

Recently, at the end of 2022, litigation started over hair relaxers, alleging that people who have their hair straightened frequently, that the chemicals that are put on their scalp, that they get into the bloodstream and can lead to uterine cancer. The scientific literature around these hair relaxers is pretty small, probably not strong enough to support litigation, but the plaintiffs’ attorneys have put in their complaints about the ingredients that go into the hair relaxers, and the number one ingredient that they highlight is phthalates.

What the plaintiffs are arguing is that it's the phthalates in the hair relaxers that cause the uterine cancer. The science is a little bit stronger about that than it is about the hair relaxers themselves.

Insurance Thought Leadership

To probe a little bit on that. When you spot something like this, insurance companies are buying this information from you, and they're changing how they underwrite? Or what exactly happens once you identify a problem like this potential one?

Robert Reville

When an insurance company works with us, the first thing they normally do is they take their casualty book of business and they enter it into our system. We model it, and we identify the exposure to all of the emerging risks in their portfolio. 

A typical situation would be, let's say, a particular company runs their model, and we find that the number one emerging risk in their portfolio is phthalates. At that point, they'd say, Oh, we need to control our total exposure to phthalates. So they develop an underwriting strategy that says, Well, if it's a chemical company, then we don't want to write that any more.

Then if there’s a downstream user of the chemical, such as a metal tubing manufacturer, we have a referral process. We want to evaluate the way the company handles the chemical, and then we want to control our total aggregations. 

All that data flows into an underwriting strategy, and then the underwriters at the company will follow the instructions that go with the strategy. So they have a new submission or a renewal, and they call up [the instructions]. They see that there's exposure to phthalates. There's either an exclusion required or a referral, and then when [a policy] is actually written, it goes into the exposure management system to control the total aggregation. That's an example.

Insurance Thought Leadership

Okay. So I cut you off. You were saying….

Robert Reville

Well, I was about to say that the question on phthalates in hair relaxers is, if the science has accepted that the phthalates are causing the uterine cancer, then what we expect is that it'll spread from hair relaxers to all these other uses. Then we'll have a full-on, everywhere chemical litigation that's happening at the same time we have the forever chemical litigation. That's potentially a worrisome situation for casualty insurance. 

Insurance Thought Leadership

Okay. And then microplastics?

Robert Reville

Microplastics are the breakdown particles as plastics degrade in the environment. And, as I mentioned, we track science from the time of the early studies as an early warning system. The literature around microplastics has grown faster than any literature we have been tracking since the beginning of Praedicat. There are now about 15,000 studies published that are examining the impact of microplastics in the environment, the way in which microplastics can enter the body, the way in which they're carried by weather around the globe, where they go once they enter the body, how they end up in the brain. 

Recently, there was a study that actually was looking at how microplastics, which are in the atmosphere primarily in urban areas because of tires -- they come off of tires – actually attract water vapor at different temperatures than clouds normally form at. They start to form clouds and can change weather patterns.

The extent of impact of microplastics in the environment is so great that the potential avenues for litigation being identified in the scientific literature are pretty enormous. We think the impact could come out in many different ways and are working with our clients around exactly how to think about where the exposures might be. There has been some litigation, nothing insurance-relevant yet, but consumer litigation and the like. We are seeing 65 cases in the U.S. courts right now…. We can start to get a sense of who the defendants might be and start to work with our clients early on what their total exposure might be.

Insurance Thought Leadership

That's fascinating. Any other things we ought to be watching for?

Robert Reville

Well, we have 300 in our system. 

I would say that another area that is worth talking about is generally climate. Climate is an area where there has been some litigation. In particular, there are about 50 cases in the U.S. courts, typically brought by states’ attorneys general, and they are alleging that, typically, oil and gas companies have caused climate change. Those cases are really the first wave of potential litigation over climate. And when we say climate, we expect there's going to be potentially litigation over what caused climate change. That may also be followed by litigation over failure to protect against the consequences of climate change. We've already seen some of this. 

I'll give you a quick example, though, of the kinds of unexpected things that will happen as a result of climate change. The scientific literature, in the mid-2010s, about 2015 or 2016, started to identify this issue that increasing droughts and increasing heat were causing concentrations of arsenic in soil. Rice plants tend to pick up the arsenic from the soil, and so some scientists started to argue that because of the fact that baby food uses rice as its backbone, that there would be litigation over elevated levels of arsenic in rice in baby food. We actually caught that literature and highlighted it to our clients before the litigation began. Then in 2021, a congressional report came out looking at heavy metals in baby food, specifically identifying elevated levels of arsenic, and the litigation was off to the races. 

It's still going on. There are, I think, both baby food manufacturers and retailers named as defendants. There are, I think, over 100 defendants so far, thousands of cases, and we'll see where it goes. But the interesting thing is all that was caused essentially by the changes in the environment from climate change.

You could argue that's the first product liability that resulted from climate change. Those kinds of things, we expect to see more of. 

We're also doing a lot of work on potential liability consequences after urban heat waves. Scientists have argued that in coming years the potential is for there to be heat waves that are worse than anything we've ever seen. If combined with blackouts, there could be a large amount of deaths in urban areas. Given those types of models, we have looked at the potential liability consequences for companies that would arguably have been responsible for taking care of vulnerable parties. So nursing homes, schools, day cares, etc., also habitational real estate. 

This is a phenomenon that hasn't yet happened, but it's worth working through the liability consequences, given the projections coming from the climate scientists. 

Insurance Thought Leadership

Great. A bunch more reasons for me not to sleep at night. 

Thanks for taking the time. Congratulations on being a finalist, and I'll be fascinated to follow your work.

Robert Reville

My pleasure.