Twenty years in insurance technology taught me one hard lesson: traditional desk underwriting fails to address the economics of small accounts. From building a data analytics and predictive modeling company to scaling a company that specializes mainly in workers' compensation insurance for small businesses, my career tracks the evolution of data and automation and the progression from InsurTech 1.0 to 2.0.
My next step: Building out an AI-native MGA. Utilizing an AI-native architecture makes small artisan contractor E&S accounts profitable for every stakeholder in the value chain.
Act One: A better compass on the same ship
I co-founded Valen Analytics in the mid-2000s because property and casualty underwriting relied on actuarial tables describing outdated market conditions. Desk underwriters assessed accounts with limited information.
We introduced predictive modeling as a solution on top of existing carrier systems to give underwriters the ability to price accurately.
Valen succeeded because it brought objectivity to a subjective process. We proved that predictive analytics drive underwriting profitability across multiple lines.
However, Valen functioned as a layer on top of existing desk underwriting processes. It improved human decision-making but failed to alter the underlying carrier operating model. We built a better compass. The ship remained a legacy vessel.
Act Two: We rebuilt the platform and found the next wall
At Pie Insurance, we remade the small business insurance experience. We automated workers' compensation pricing and underwriting with data science, at an exceptional loss ratio. Growing to more than 38 states, 5,000 agencies, and more than $500 million in premium, Pie represents the peak of the Insurtech 1.0 wave.
However, modern insurance carriers face a structural reality: Scaling operations traditionally requires adding staff. Underwriting assistants chase down missing forms; operations teams process endorsements; and specialists handle cancellations and reinstatements. It's a labor-intensive model that causes unit economics to worsen for smaller accounts. The fixed cost of servicing a policy remains high regardless of premium size. This operational wall limits efficiency.
Act Three: AI-native means the brain is the product
Significant advancements in AI enable this third act, which is a significant departure from Insurtech 1.0. AI-native architecture builds from the ground up on an agentic foundation, not as a bolt on to a legacy stack.
An AI-native system acts as an underwriting brain. It processes artisan contractor accounts with machine-level consistency. The system enables us to solve the unit economic problem of small-premium contractors. A Decision Engine treats a $3,000 account with the same underwriting rigor as a $300,000 opportunity. Because the marginal cost of this rigor is effectively zero, we can deliver profitable underwriting for small E&S premiums.
InsurTech 2.0 — The Power of agentic underwriting
Agentic AI will define the next decade of insurance. Traditional automation relies on static "if-then" logic. Agentic systems operate with autonomy, reasoning, and intelligence. These systems ingest complex submissions, evaluate risk depth, and manage binding authority through dynamic algorithmic parameters. Exceptions are surfaced for underwriter review (i.e. "the human in the loop").
For the wholesale broker, this means a clean, fast Quote-Bind-Issue experience on accounts that have historically been economically painful to service. The broker's economics change when the MGA/carrier runs on agentic infrastructure: Submissions get quoted or declined in minutes, endorsements do not sit, and binders process without manual intervention. For a segment as operationally intensive as small E&S artisan contractors, this distinction is the difference between a viable book and an unprofitable one.
This architecture shifts insurance professionals from roles as execution engines to strategic pilots. Intelligence and autonomy scale independently of headcount. Underwriters spend their time developing the system, not data entry and research. This model provides the foundation for sustainability in the Excess and Surplus (E&S) market. Agentic AI maintains underwriting rigor while eliminating the variable costs of manual labor.
Building for the future
The current E&S market requires speed, precision, and data-anchored results. Legacy technology debt limits the ability of traditional MGAs and carriers to leverage data effectively. It reduces profit potential and slows product innovation. And it keeps the industry from serving the needs of policyholders who end up not getting the coverage they need at a price they can afford.
AI-native platforms eliminate this friction. Specialized solutions now exist for the SMB artisan contractor market. These platforms use simplified data sets to drive rapid deployment. Underwriting rigor remains the durable advantage of the agentic era.
Dax Craig is Co-Founder and CEO of Comeryx, an AI-Native MGA serving the small artisan contractor market. Previously, he was the co-founder and President of Pie Insurance, and co-founder and CEO of Valen Analytics. Dax has over two decades of experience driving strategic growth at the intersection of insurance and technology.
Opinions expressed here are the author's own.
(Featured image credit: Johanna Pung/Wikimedia Deutschland)
© Arc, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.