Why Carriers Need AI for Pricing
Pricing is the most consequential underwriting decision a carrier makes. Inadequate rates are the primary driver of combined ratio deterioration, and rate inadequacy that persists for multiple policy terms can take years of corrective action to unwind. The carriers with the most consistent underwriting profitability are not simply better at predicting individual risks -- they are better at identifying segments where their current pricing is inadequate and adjusting before the losses confirm what the data already suggested.
Traditional ratemaking is actuarially sound but operationally slow. Annual rate reviews, GLM-based models, and regulatory filing timelines create a cycle that struggles to keep pace with the frequency of external changes affecting loss costs -- medical inflation, litigation environment, climate exposure, supply chain disruption. AI pricing tools do not replace actuarial methodology; they accelerate it, allowing actuarial teams to identify pricing signals, test model updates, and deploy rate changes faster than the traditional cycle permits.
For foundational context on how predictive underwriting and pricing intersect, see the AI underwriting glossary entry. For a broader view of the carrier AI toolset, see the companion AI underwriting tools for carriers page.
Key Use Cases and Workflow
Personal lines rate plan modernization. The highest-volume use case for AI pricing tools is modernizing personal lines rate plans -- auto, homeowners, renters -- from GLM-based models to gradient boosting or neural network approaches that can incorporate more variables and identify micro-segments where current pricing is inadequate. Platforms like Akur8 are specifically designed for this use case, with actuarial interpretability built into the model outputs so that actuarial teams can validate results against their own judgment.
Small commercial pricing refresh. Small commercial lines have historically been priced off simplified rating algorithms because the per-account premium does not justify extensive individual underwriting. AI tools that score small commercial risks against industry and location signals allow carriers to differentiate pricing within the small commercial segment without adding underwriter headcount. The actuarial cycle time improvement is particularly valuable here -- small commercial pricing often goes years between meaningful updates in traditional environments.
Specialty lines pricing support. For complex and specialty commercial risks, tools like Hyperexponential support underwriter-led pricing decisions with ML model guidance. The underwriter retains judgment on individual risks, but the platform provides data-driven pricing signals calibrated against the carrier's historical book and external benchmarks. This is the use case where human judgment and AI tools are most tightly integrated rather than operating sequentially.
Climate and property risk scoring. ZestyAI uses aerial imagery and environmental data to score property-level climate risk -- wildfire, flood, severe wind -- at a granularity that standard rate plan variables cannot achieve. Carriers using ZestyAI integrate property risk scores into their homeowners and commercial property rate plans to improve accuracy in high-exposure geographies. The platform is particularly relevant for carriers with significant homeowners exposure in California, Gulf Coast states, and the Midwest hail corridor.
Predictive loss modeling for pricing. Tools like Pinpoint Predictive use non-traditional data signals to predict loss propensity at the policyholder level. The use case in pricing is identifying, at new business or renewal, which accounts are likely to be unprofitable even when standard rating variables suggest otherwise. This is most valuable in personal lines, where the premium per account is too small to support individual underwriting review.
Pricing validation and audit. A less commonly discussed but practically important use case is using AI pricing tools to audit existing rate adequacy -- running the current book through a new model to identify segments where the carrier is systematically under-priced or over-priced. This diagnostic use case can justify the technology investment even before the carrier is ready to deploy the new model in production.
What to Look For
Actuarial methodology compatibility. State regulators require rate filings to be supported by actuarial certification. AI pricing tools that produce opaque models which actuaries cannot validate or certify will create regulatory filing problems. Look for platforms that support actuarial review workflows and produce model documentation in formats that state insurance departments can evaluate. Platforms built to work alongside actuarial teams -- rather than replace them -- have better track records of regulatory approval.
Integration with policy admin and rating engine. An AI pricing model that produces recommended rates but cannot connect to the rating engine creates a manual process that undermines the speed advantage. Evaluate integration depth with your specific policy administration system and rating workflow.
Regulatory filing requirements. Some states require pre-approval of rate changes; others allow use-and-file. The speed advantage of AI pricing tools is partially constrained by state filing timelines. Understanding your specific regulatory environment is necessary before modeling the operational benefit of faster rate cycles. Carriers writing in large states like California, New York, and Florida face more filing complexity than those concentrated in use-and-file states.
Actuarial team adoption. The change management challenge in AI pricing is as significant as the technology challenge. Actuarial teams that have built and maintained GLM models for years may be skeptical of ML approaches they did not develop. Vendors with strong actuarial team onboarding programs and interpretability tools have higher adoption rates than those that treat actuaries as an obstacle to navigate rather than a constituency to serve. Before selecting a platform, ask vendors for references from actuarial departments -- not just IT or operations -- at comparable carriers. The actuarial team's willingness to use the platform in their daily workflow determines whether the investment produces ROI or sits underutilized after go-live.
Total cost of ownership. AI pricing platform costs include license fees, implementation, actuarial training, and ongoing model maintenance. See the total cost of ownership glossary entry for a framework applicable to insurance technology evaluation generally. The annual maintenance and model refresh cost is often underestimated relative to the initial implementation cost.
Akur8
Akur8 is an actuarial pricing platform that uses ML to augment traditional ratemaking. The platform uses gradient boosting to identify loss ratio patterns by segment and produces models that are interpretable through actuarial methods -- a critical requirement for state rate filings. It is strongest in personal lines pricing modernization, where carriers are moving from legacy GLM models to more granular, more frequently updated rate plans. Actuarial teams use Akur8 alongside their existing ratemaking workflow rather than replacing it. Pricing is quote-based.
For a direct comparison, see Akur8 vs. Hyperexponential. The two platforms serve different primary use cases -- personal lines actuarial modernization vs. specialty lines underwriter pricing support -- and are rarely direct competitors.
Earnix
Earnix is an enterprise rating and pricing personalization platform for large carriers. It covers pricing model development, rating engine deployment, and customer-level pricing optimization -- a broader scope than pure actuarial modeling tools. Earnix is used by large carriers that want a single platform for pricing across multiple lines and distribution channels. The pricing personalization capability -- adjusting offer prices at the customer level based on retention propensity, channel, and willingness-to-pay signals -- extends beyond what a pure actuarial modeling platform provides. Carriers that have completed initial rate plan modernization and are looking to optimize pricing at the customer interaction level, not just the segment level, find Earnix more relevant than actuarial-only tools. Pricing is quote-based.
Hyperexponential (hx Renew)
Hyperexponential is a pricing platform for complex commercial and specialty lines. It is designed for underwriter-led pricing on excess, casualty, specialty, and London market risks -- lines where individual underwriter judgment is central and where a pure algorithmic model is insufficient. The platform provides ML-based pricing guidance that the underwriter can accept, adjust, or override, with full audit trail. It is one of the stronger options for specialty and excess carriers writing complex risks where pricing requires judgment that models alone cannot replicate. Pricing is quote-based.
ZestyAI
ZestyAI integrates aerial imagery, environmental data, and climate models to produce property-level risk scores for homeowners and commercial property lines. The platform is used by carriers to refine property pricing in high-exposure geographies -- particularly wildfire and flood zones -- where standard rating factors do not capture true location-level risk. ZestyAI scores can be integrated into rate plans to create more granular property pricing without requiring full actuarial model rebuilds. Pricing is quote-based.
Pinpoint Predictive
Pinpoint Predictive uses predictive analytics and non-traditional data signals to identify loss propensity at the policyholder level. Its use case in pricing is identifying at-risk segments that standard rating variables underweight -- helping carriers improve rate adequacy in segments where loss experience has exceeded expectations. The platform's approach is to identify signals in data carriers already collect -- or that can be sourced from third parties -- that are predictive of adverse loss development but not currently reflected in rating plans. For personal lines carriers facing adverse selection in specific renewal cohorts or geographic segments, Pinpoint Predictive can surface the data pattern driving that selection even when standard rating factors appear adequate on their face. Pricing is quote-based.
