Insurtech Spring Has Sprung
By: Teddy Himler | March, 2025
By: Teddy Himler | March, 2025
How AI Can Fuel an Insurtech Renaissance
At Optimist Ventures, we invest in industry-transforming technology companies in the largest labor sectors. We are particularly excited about the transformations afoot in the $9T global insurance industry, driven primarily by AI.
Insurance isn’t just a large market; it’s one of the most structurally inefficient and least digitized, despite being one of the most data-rich and manual. That’s exactly why we believe it is poised for generational change driven by AI and other transformational technologies.
Insurers are relatively underinvested in R&D, have aging workforces, swim in oceans of unstructured data, and suffer from deeply manual processes — this makes the sector ripe for AI-driven optimization.
Consider, only 9 insurers rank in the world’s top 2,500 R&D spenders. While R&D expenditures in the industry are trending up annually, P&C and Life insurers reinvest just 5% of written premiums into IT (see below) — significantly below the 8% of revenue banking sector benchmark.
Moreover, while the industry generates over 2.5 billion gigabytes of data daily, most of it flows through legacy software solutions like Guidewire, Duck Creek, or Vertafore. Cloud, mobile, and SaaS waves largely passed by insurance with limited effect. That makes this not just a late adopter industry — but potentially the best-positioned to leapfrog via AI-native infrastructure. (Below: Insurance industry’s R&D expenditure, as % of premiums)
How AI Can Fuel an Insurtech Renaissance
At Optimist Ventures, we invest in industry-transforming technology companies in the largest labor sectors. We are particularly excited about the transformations afoot in the $9T global insurance industry, driven primarily by AI.
Insurance isn’t just a large market; it’s one of the most structurally inefficient and least digitized, despite being one of the most data-rich and manual. That’s exactly why we believe it is poised for generational change driven by AI and other transformational technologies.
Insurers are relatively underinvested in R&D, have aging workforces, swim in oceans of unstructured data, and suffer from deeply manual processes — this makes the sector ripe for AI-driven optimization.
Consider, only 9 insurers rank in the world’s top 2,500 R&D spenders. While R&D expenditures in the industry are trending up annually, P&C and Life insurers reinvest just 5% of written premiums into IT (see below) — significantly below the 8% of revenue banking sector benchmark.
Moreover, while the industry generates over 2.5 billion gigabytes of data daily, most of it flows through legacy software solutions like Guidewire, Duck Creek, or Vertafore. Cloud, mobile, and SaaS waves largely passed by insurance with limited effect. That makes this not just a late adopter industry — but potentially the best-positioned to leapfrog via AI-native infrastructure. (Below: Insurance industry’s R&D expenditure, as % of premiums)
But Why Now?
Insurance is the arena in which technology’s inexorable march — new applications, new models, new sensors, new risks — is squaring off against turbulent, generational macro forces in demographics, climate, and geopolitics.
Despite these challenging circumstances, we are optimistic.
Unlike past Netflixing or Uberized disruption, we believe insurtech is a key enabler for incumbents to navigate these gale-force macro headwinds, a tool for flexibility and survival.
As of 2024, 77% of leading insurers have begun investing in AI implementation, bearing early fruit. In a pilot outfitting human claims agents with AI email-writing co-pilots, Allstate found AI-generated correspondences to be consistently “more empathetic and less accusatory” than human-written messages. Sapiens has been developing its GenAI toolkit in partnership with Microsoft and OpenAI for over 2 years. Emerging players, such as 9Root and Outmarket, are raising fresh capital to design end-to-end human/AI automation processes for large insurers. Progressive, perennially a technology leader, has achieved 5x claims adjuster productivity gains at scale, attributable to computer vision photo estimates and task automation.
Where the financial industry stood 10 years ago before the broader Fintech boom, we today see insurance at the same inflection point. Embracing AI is critical for insurance’s continued evolution, and we are excited for the industry to “leapfrog” others in embracing tech and course correcting.
Our Learnings from Insurtech 1.0
Our prior work as meaningful early investors in Hippo and Lemonade provides great insights into the future. Starting in 2015, insurtech surged, peaking in 2021 with a $9B valuation for Lemonade (when it had ~$200M of In-Force Premium and ~$63M in adjusted gross profit). Investors were lured by massive TAM assumptions, wonky valuation methods, a “disruptive” D2C playbook, and aggressive fundraising.
Soon came the insurtech retrenchment. Interest rate hikes, tech-investing sobriety and a broader macro correction dashed 2021’s $17B investment and insurtech darlings Lemonade, Hippo, and Root saw revenue multiples continue to decline, the space entered an accelerated correction (leaving us with insurtech’s lowest-ever annual funding since 2018 — see below).
Despite these challenging circumstances, we are optimistic.
Insurtech 2.0 is Different
The faltering of Insurtech 1.0 wasn’t about tech adoption, but rather challenged business models and misunderstood dynamics. Some lessons we’ve taken to heart:
Insurance-first, not tech-first. Underwriting discipline and risk judgment are not optional. While early insurtechs thought of themselves as tech-first companies, domain expertise was overlooked by industry outsiders unfamiliar with the insurance landscape. Just as one should be wary of the “creative” accountant, having an insurer who “moves fast and breaks things” is contrary to the end-consumer’s best interest. Lemonade infamously boasted of its AI-driven, lean adjusting team until its Q4’21 loss ratios reached 96%, forcing it to hire seasoned human adjusters.
Loss ratios > logos. Nail loss-ratio essentials, then scale. While hyperscaling may work in the B2B SaaS playbook, this is a red flag in insurers who operate in efficient markets, and over-binding policies too quickly indicates a lapse in the insurer’s judgement. Somewhere, there is a coverage gap or the market is taking advantage of an underwriting inefficiency which, with time, will develop into an unfavorable loss ratio.
Avoid Competition. Going head to head with State-Farm, Progressive, Allstate makes capital efficiency very difficult. Homeowners, renters, personal auto, etc. are commoditized, efficiently priced, and distributed by incumbents with vast resources. UX is not sufficient to run a lean customer acquisition strategy. The esoteric specialty lines and new and emerging risk categories offer a wonderful complex world of opportunity
The agent isn’t going anywhere fast — legacy distribution remains sticky. Humans crave human interaction, especially when purchasing high-trust products. Distribution through intermediary agents will remain resilient, and new tech will only increase agent-broker throughput rather than disintermediate the value chain.
Now What?
Sobered by the learnings of Insurtech 1.0, though, we see a gradual thawing of the insurtech winter, and are now leaning in when others have evacuated the market.
While industry-wide cyclical funding is down, AI-related funding now constitutes ~2/3 of all insurtech deals (compared to 12% only 9 months before). Meanwhile, M&A dealflow in the space has remained resilient despite a dismal outlook in most other sectors. To list just the most recent activity, Munich Re announced its majority stake in Next Insurance at a $2.6B valuation in March 2025, only weeks after Vouch announced its acquisition of StartSure.
With Ethos exploring a potential IPO this year, we anticipate coming quarters to see a surge in private market insurtech funding, with investor confidence bolstered by quietly resilient publicly-traded insurtech performance. In its Q4’24 earnings call, Lemonade (LMND) reported 24% YoY growth in in-force premiums (IFP) and best-ever loss ratio of 73%, in part attributable to technology and part process. Root (ROOT), reaching profitability for the first time and their stock appreciated 3x+ YoY.
Insurtech is on the up again — this time built on more durable foundations.
What we are looking for in Insurtech 2.0
In many ways, it resembles fintech ~2014: infra-first, painkiller-oriented, and deeply aligned with industry incumbents. Insurtech 2.0 isn’t about displacing the value chain; it’s about extending and enhancing it using AI-native tools that are interoperable with incumbents and capital efficient. 2.0 companies will be lean, ship fast, and scale revenue well ahead of headcount.
While still in the megatrend’s infancy, hallmarks of AI-native companies are emerging: these are lean (often sub-20 employees) scaling to millions in ARR just months after launching. Think of Mercor (30 FTEs) at $75M of ARR or Lovable (15 FTEs) at $17M of ARR. Small teams quietly scaling, iterating products in tighter feedback loops, and leveraging real AI-driven ROI for unprecedented revenue efficiency. We expect insurtech companies of this profile to materialize this year, guiding Optimist Ventures’ early backing of Harper.
Four company archetypes powering the next generation
Emerging risk specialists | From AI liability to wildfire exposure and autonomous vehicles, these companies specialize in modeling, pricing, and underwriting new and poorly served risks. A first-mover profile affords them data moats driving AI risk models, but are prone to vulnerabilities involved in taking incumbents head-on. Opportunity lies in eventual acquisition or partnership for access to their niche underwriting skillset or tech stack. Look out for companies building in the parametric space as well, with predictive weather modeling (see Microsoft’s quiet progress on Aurora) advancing in leaps and bounds and US municipalities leaning into these solutions.
Examples: Shepherd (Construction), Koop (Frontier Tech), Charter (Space), Floodbase and Kettle (Climate), Mayflower and Testudo (AI)
Insurance ontologies | Think Palantir for Insurance: Ontologies and knowledge graphs that structure fragmented enterprise data, and enable agentic workflows. Palantir modeled a per-customer ontology consulting business for governments’ complex, repetitive workflows in environments with siloed data. Sounds like the perfect approach for insurance.
AI co-pilots and automation layers | Off-the-shelf agents for underwriting, claims, customer comms, and fraud. These solutions are high-repeatability, plug-and-play, and built to coexist with existing platforms — accelerating throughput without deep displacement. These are things like email co-pilots, AI voice-based customer service representatives or receptionists, and simple auto-complete solutions. They must be integrated with existing, typically legacy, systems to ingest data needed to fuel LLM-driven personalized recommendations, claims filing, or ustomer service.
AI-native distribution infrastructure | Finally, API-first engines for quoting, binding, and administering policies with radically reduced operational overhead. These players enable next-gen speed and UX across incumbent distribution channels.
Conclusion: The Slope of Insurtech Enlightenment
Like many in the tech industry, we are believers in Carlota Perez’s framework of technology revolutions. Often early, “primitive” forms of an emerging technology emerge and capture the imagination and unbridled enthusiasm of capital markets in the installation periods. Creative valuation techniques are applied to the new new thing (e.g., GWP multiple vs. DCF). These solutions are imperfect, but exuberance masks shortcomings. Soon, the bubble pops and the market is evacuated. After a trough of disillusionment, deployment occurs and durable businesses are created.
While still in the megatrend’s infancy, hallmarks of AI-native companies are emerging: these are lean (often sub-20 employees) scaling to millions in ARR just months after launching. Think of Mercor (30 FTEs) at $75M of ARR or Lovable (15 FTEs) at $17M of ARR. Small teams quietly scaling, iterating products in tighter feedback loops, and leveraging real AI-driven ROI for unprecedented revenue efficiency. We expect insurtech companies of this profile to materialize this year, guiding Optimist Ventures’ early backing of Harper.
Four company archetypes powering the next generation
Emerging risk specialists | From AI liability to wildfire exposure and autonomous vehicles, these companies specialize in modeling, pricing, and underwriting new and poorly served risks. A first-mover profile affords them data moats driving AI risk models, but are prone to vulnerabilities involved in taking incumbents head-on. Opportunity lies in eventual acquisition or partnership for access to their niche underwriting skillset or tech stack. Look out for companies building in the parametric space as well, with predictive weather modeling (see Microsoft’s quiet progress on Aurora) advancing in leaps and bounds and US municipalities leaning into these solutions.
Examples: Shepherd (Construction), Koop (Frontier Tech), Charter (Space), Floodbase and Kettle (Climate), Mayflower and Testudo (AI)
Insurance ontologies | Think Palantir for Insurance: Ontologies and knowledge graphs that structure fragmented enterprise data, and enable agentic workflows. Palantir modeled a per-customer ontology consulting business for governments’ complex, repetitive workflows in environments with siloed data. Sounds like the perfect approach for insurance.
AI co-pilots and automation layers | Off-the-shelf agents for underwriting, claims, customer comms, and fraud. These solutions are high-repeatability, plug-and-play, and built to coexist with existing platforms — accelerating throughput without deep displacement. These are things like email co-pilots, AI voice-based customer service representatives or receptionists, and simple auto-complete solutions. They must be integrated with existing, typically legacy, systems to ingest data needed to fuel LLM-driven personalized recommendations, claims filing, or ustomer service.
AI-native distribution infrastructure | Finally, API-first engines for quoting, binding, and administering policies with radically reduced operational overhead. These players enable next-gen speed and UX across incumbent distribution channels.
Conclusion: The Slope of Insurtech Enlightenment
Like many in the tech industry, we are believers in Carlota Perez’s framework of technology revolutions. Often early, “primitive” forms of an emerging technology emerge and capture the imagination and unbridled enthusiasm of capital markets in the installation periods. Creative valuation techniques are applied to the new new thing (e.g., GWP multiple vs. DCF). These solutions are imperfect, but exuberance masks shortcomings. Soon, the bubble pops and the market is evacuated. After a trough of disillusionment, deployment occurs and durable businesses are created.
As suggested, the paradigm shift of cloud and mobile create a $58B frenzy for insurtech from 2017–2021. Today, the innovators have learned the right lessons, and are armed with a new set of tools, particularly AI. We at Optimist Ventures believe that we are entering into a more substantive insurtech future, and it couldn’t come at a better time.
See here our Q1'25 Insurtech Spring Market Map
