Modernizing the Insurance Industry

Insurance is a backbone to our economy, allowing people and businesses to take risks by limiting downside protection. Coupled with financial services, the finance and insurance sector accounted for over 7% of US GDP. The insurance industry employs about 2.9 million workers which represents over a quarter of the workforce in the financial services sector. In 2023, overall Gross Written Premiums (GWP) in the US was $2.1T and projected to reach over $3T in 2027, driven by rising social inflation and reinsurance rates.

Over the years, the insurance industry has undergone multiple transformative waves of innovation since its formalization, each driven by technological advancements, regulatory shifts, and evolving consumer needs. From the establishment of foundational companies like Traveler’s in the 19th century to the rise of artificial intelligence in the the 21st century, each phase reflects the sector’s adaptability and resilience. 

Throughout time, carriers have introduced new insurance products that encompass new risks. From the early innings of auto insurance to workers compensation to P&C starting with fire insurance, new types began to emerge with modern risks. At Montage, we’ve been active investors in insurance across multiple fund lifecycles. We’ve invested across innovative insurance lines like Vouch which fulfills SMB business insurance needs and Future Family which helps families afford fertility financing and IVF insurance. We’ve also made infrastructure investments that drive modern workflows, including claims management automation providers like Cape Analytics, Snapsheet, and Sure which helps partners embed insurance products.

The next wave of innovation in insurance will not only redefine how risk is managed but also unlock new, previously uncharted markets. Below are sectors of insurance we’re closely tracking and looking for opportunities in.

New Specialty Insurance Products  

The E&S market has seen substantial growth, reaching $115 billion in direct written premiums in 2023, a 16.8% increase from the prior year. E&S lines now constitute 9.2% of the total U.S. direct premiums written in 2023. E&S insurers are developing more tailored solutions to address evolving and complex risks that traditional insurers are hesitant to underwrite. With the rise of new industries, technologies, and business models, there's a constant demand for innovative insurance solutions. The E&S market is often at the forefront of developing products for new and emerging risks, as we’ve seen with examples like cyber insurance, errors & omissions for construction, specialty lines for supply chain, or property insurance for high risk areas of natural disasters. 

New Insurance products for AI

As new AI applications will be developed and models will be relied on for decision making processes, there are greater risks associated with performance warranties for AI models. Unlike cyber insurance which typically covers operational losses from cyber attacks and data breaches, AI insurance will be focused on potential liabilities and risks from AI products that are recommending and interacting with customers. 

Last year in February 2024, Air Canada was sued and ordered to pay damages after its AI chatbot gave a customer incorrect information about bereavement fares (source)  Air Canada paid the customer the equivalent of the difference between what he paid for his flight and a discounted bereavement fare, alongside additional interest and fees. While it was a nominal amount for Air Canada, companies leveraging AI solutions and agents for decision making will need to protect against potential errors that can lead to financial loss, discrimination, or harm at scale. In addition, as states begin to pass laws regulating the use of AI in insurance, having appropriate coverage can help companies navigate evolving regulatory landscapes. One recent ecample includes Munich Re's new offering for startups to receive aiSure to cover for AI models and performance.

Startup Examples: Coalition (cyber insurance), Parsyl (supply chain insurance)

Addressing Climate Risk 

Climate insurance has remained a visible gap in coverage to protect individuals, businesses, and governments against financial losses by extreme climate related events. Verisk reports that the average annual cost from global natural catastrophes has reached a new high of $151B, as the increasing event frequency and magnitude of losses continues to climb. This challenge has driven sky high premiums and the retraction of insurers from high-risk locations like California and Florida, forcing many property owners to rely on state-backed insurance options or self-insure.

Outside of direct weather events, there are even greater consequences including supply chain disruptions and replacement costs. The pressure to consider climate disasters in business planning is also increasing as regulatory bodies like the European Banking Authority have published guidelines to quantify ESG risks and the Critical Entities Resilience Directive (CER) has entered into legislation this year.

As a result, we’ve seen new forms of parametric insurance built for specific climate related risks, such as helping auto dealers and property managers protect against hail (Understory), new forms of property insurance (Stand), or compliance workflows with these regulatory tailwinds (Eoliann).

However, when addressing climate risk, there are also many picks and shovels to be built in having the right data infrastructure to forecast climate events. One example is real time weather forecasting that can better ingest sensor data across the world that has previously been underutilized, improving legacy numerical weather prediction models that were conceived in the 1950s. New large physics models can significantly improve the way that insurers can forecast and quantify climate risk in the future.

Examples: Plover Parametrics (specialty and climate brokerage), Brightband (AI weather forecasting).

Infrastructure Workflows

AI Powered Client interactions

While digital distribution channels have evolved in reaching consumers, the share of premiums through agents and brokers continues to dominate, particularly in the P&C and commercial lines. There are approximately over 421k insurance brokers and agencies in the US and roughly 40k being independent P&C agents and brokers. Total P&C direct written premiums reached $952 billion in 2023 in the US and the independent agency channel placed 87% of commercial lines premiums and 39% of personal lines (Insurance Journal).

Consequently, trust continues to be a driver for independent agents to be advisors for the clients. In fact, insurance remains the leading industry for outbound call volume behind lending services. Customers value human interaction for insurance experiences from the sales to the claims experience. However, customers have also made it clear that they value access to self-serve, digital options alongside flexible communication across channels. 

Looking forward, we’re excited about purpose built agents for the insurance sector that provide tailored, personalized support across the lifecycle of acquisition to retention and support. Insurance is a unique vertical that requires nuanced understanding of workflows and read/write into critical sources like AMS and CRMs. 

Examples: Regal (Customer service across industries), Domu (voice calls for lending and insurance)

Intelligent Underwriting & Decisioning

A recent Accenture survey found that underwriters spend 40% of their time on administrative tasks, amounting to two days of their work week. Inboxes are flooded with more submissions than ever, but by some estimates, underwriters are only able to respond to 10%. Typical workflows include underwriters re-keying data from policy admin systems and modeling in excel spreadsheets, where parameter changes can lengthen the process. Many underwriters and actuaries are still modeling in their own disparate excel sheets, homegrown systems, or pay for consultancies’ services. 

One major opportunity in underwriting is enhancing the decisioning and risk intelligence process to 1) manage existing risk in the portfolio and renewals, and 2) price new risk and accounts. The unlock of AI is ingesting a variety of structured and unstructured data across datasources, leading to better risk assessments, decisioning, and profitability. For instance, AI driven underwriting models can enable granular rates based on real time data, demand, and policies based on individual risk profiles. In addition, AI can integrate historical and real time claims data to better understand impacts on loss ratio and payouts, improving underwriting models. While AI will significantly enhance underwriting capabilities, we expect it to complement rather than replace human expertise in the industry.

Most insurers, despite being businesses built on underwriting and pricing data, are still in the early stages of capturing the full potential of modern analytics and ubiquitous data. While there have been investments on the automation of data, lead generation, quoting, rating, and underwriting, the data used in underwriting and quoting feeds does not feed to claims, and is siloed. Carriers’ claim notes typically aren’t shared to underwriting or vice versa. More sophisticated insurers have homegrown claims systems that provide a feedback loop to the actuarial and underwriting teams for them to monitor the overall portfolio daily, compare performance across time frames, and help price at the renewal. Talent remains a challenge as many claims adjusters have built out their own industry knowledge for a decade and are now retiring, or folks are leaving the industry for more lucrative jobs. New platforms that can integrate claims, underwriting, and actuarial data can help train new talent entering the workforce. 

Examples: Sixfold (automated insurance underwriting), Federato (risk operations), Hyperexponential (underwriting for complex risks)

Claims Automation

Approximately 68% of US premiums collected for property and casualty was spent on paid losses and claims in 2023. The components of claims include 1) intake and triage, 2) core claims handling, 3) specialized claims handling, 4) payments. The process involves receiving the first notice of loss, verifying the claim’s policy and coverage, communicating to the insurer any recommended actions, predicting estimated losses, verifying any potential complications (i.e. litigations), and making payments or reimbursements. The average claim can be resolved in days to several months, depending on the type of claim and complexity. While auto claims and simple property claims can be resolved quickly with virtual appraisals or aerial imaging, others like workers compensation, homeowners insurance, and commercial claims can take up to several months.

Within claims automation, we’ve seen some companies tackle segments including intelligent triaging to prioritize claims, 24/7 customer support through chatbots, or enhanced fraud detection. However, given the complexity of the multi-step workflow for more complex claims, we’re excited about AI agents that can serve adjusters throughout the entire case – document preparation, policy verification, fraud checks, analysis of outcomes, legal strategy and negotiation. The ability of generative AI to help analyze vast amounts of data, generate claim summaries with actionable insights, leverage predictive models, and orchestrate each step at scale will be a big unlock of time savings, research, and accuracy. 

Examples: Assured (P&C claims), Snapsheet (virtual appraisals for auto), Clara Analytics (claims triage and benchmarking)

New Data Models and Providers

Specialty insurers are still struggling to find the right datasets, whereas typical lines like auto, homeowners insurance, and small commercial typically use the same 10-15 data elements to write insurance for the last 50 years. These lines have a plethora of databases and historicals to draw from via incumbents like LexisNexis. However, for specialty lines there isn’t as much industry, historical data, or state guidelines, leading to more leeway in pricing and potential blind spots in risk. 

The next generation platforms unseating LexisNexis as data providers can offer more granular intelligence to underwriting, decisioning, and risk management. One example is our portfolio company, Paceline, that offers proprietary, zero-party data. Their consumer fitness and social wellness platform (over 1M downloads on the app store) reads wearables to track physical activity, heart health, sleep, and personal data. They then work with life and health insurers by embedding their platform and providing an underwriting edge, reducing acquisition costs, increasing LTV, and aligning incentives with reduced underwriting risk. By starting with a unique data wedge, companies like Paceline can drive underwriting as a service. For other lines of insurance, we’ve seen interesting companies (mentioned below) that either bring in a new set of untapped data entirely to offer state of the art AI models and benchmarks.

Examples: Reask (Catastrophe risk modeling), Atmo (AI Weather forecasting), WindBorne (atmospheric data via weather balloons), Coris (SMB Risk management)

Challenges in the market

We’re looking to work with founders who recognize the opportunity and also roadblocks ahead. As we’re looking to make an investment in the insurance space, we caveat the potential concerns in the market to de-risk moving forward, including:

• Relationship driven and long sales cycles - In order for a carrier to bind, they need to do multiple dry runs and backtests. The industry is a very relationship-driven business that requires trust. Sales contracts can be longer, such as five years, resulting in deep requirements for implementation, touchpoints, and value proposition. For some startups, implementation can take months with the team building a custom model for their clients. Buyer decisions typically funnel to the CEO, leading to complex sales cycles. Generally, insurance companies allocate a small percentage of their revenue compared to other industries into technology due to cost or many tools not integrating well into underwriting. In addition, to measure efficiency and ROI, cycles in insurance can take a whole year to observe movements in loss ratios. At Montage, we work very closely with core insurance carriers and MGAs in our network that we introduce our founders to in expediting these sales cycles.‍
• Compliance -Underwriting models need to be easily explained so that it is understandable and unbiased. Regulators have accelerated their focus on potentially discriminatory underwriting practices given the industry’s use of big data and AI, as data mining might inherit any biases of prior decision makers or society at large. New AI native platforms will need to have clear data governance, explainability, and compliance with multiple levels of regulations.
• ‍Implementation - In starting with a line of business to pilot underwriting software, insurers might choose lines of business with a manageable amount of risk variables versus thousands of variables. The feedback loop is often a longer timeframe to see actual loss results, before implementing new solutions like underwriting models to lines where the bulk of premium is written (i.e. from casualty to property). Beyond the top largest insurers, the middle market and smaller insurers may be less tech savvy but move faster with appetite for tech solutions and lighter implementation. For founders building in the market, we'd be curious to hear the initial ICP and tradeoffs in serving certain customer segments.

While prior insurtech startups may have had headwinds in measuring ROI, generative AI is unlocking a new wave of innovation in the industry. For instance, historically, companies pitching a POC might invest a decade of the insurer’s historical data, take ⅔ of it to train the models, and show a better loss ratio than the other third. Startups sales pitch fell on reducing claims expenses or gaining more business by underwriting new lines. Now, with AI, ROI can be measured through concrete time savings and labor productivity in a pressing macro and labor market. We are more excited than ever about new insurance products and workflows with the unlock agentic infrastructure.

If you’re a founder building in any of these areas, we would love to chat with you and work together on paving the future. Please reach out to connie@montageventures.com.