AI is accelerating insurance pricing, but insurers still face a slower challenge: turning model outputs into decisions that can actually reach the market.
An insurance pricing model can now be built and validated in hours. Turning its output into a live price can still take months. Earnix director Mathieu Edmond argues that this gap is becoming an increasingly important competitive issue for insurers as AI accelerates the modelling process without necessarily speeding up the decisions that follow.
The shift matters because model development has traditionally been a key source of actuarial advantage. AI and automation are now making it possible to speed up modelling, documentation and analysis, but the processes surrounding those activities can remain considerably slower.
Data access, validation, deployment and governance can all add time between a model being completed and a pricing change reaching the market.
A faster model therefore does not automatically mean a faster insurer. An insurer may be able to generate and assess multiple pricing scenarios rapidly, but still need to work through internal approval processes before a new price can be implemented.
The issue becomes more complicated when pricing decisions involve competing priorities.
A model can estimate the likelihood of a loss or recommend a price based on a particular set of inputs. It cannot independently determine how that recommendation should be balanced against profitability, competitiveness, fairness and regulatory requirements.
Those trade-offs remain dependent on people and organisational processes. Pricing decisions can involve several parts of an insurer. Actuarial teams may develop and assess the model, while business lines consider its commercial impact. Distribution teams may assess how a change could affect customers and channels, while compliance and IT functions can have their own requirements before anything is deployed.
At mutual insurers, bancassurers and groups operating across multiple distribution channels, the number of stakeholders involved can make the process more complex.
The governance is there for a reason. Pricing decisions can have significant commercial and customer consequences, making oversight an important part of the process. But each additional review or approval can also extend the time between identifying a pricing opportunity and acting on it.
The result is a widening distinction between the speed of analysis and the speed of execution. Data can create another constraint. Earnix’s 2026 Industry Trends Report, The Race to Reinvent, surveyed 400 insurance executives globally and found that only 30% said their organisations can quickly access the information they need to make business decisions.
The report also found that 46% believe their technology provides the speed required for effective decision-making, while two-thirds identified poor data quality as a barrier to both decision-making and AI performance.
The findings suggest that improving model development is only one part of the challenge.
If the underlying data is difficult to access, inconsistent or unreliable, faster modelling will not necessarily result in faster decisions.
Governance presents another potential bottleneck. The Earnix research found that 92% of respondents conduct formal AI reviews at regular intervals. However, fewer than one in three executives said they were fully confident that those reviews keep pace with regulatory change.
Regulatory and legal exposure was also identified as the top ethical concern associated with AI by 38% of respondents.
For insurers, that creates a balancing act. AI can increase the speed and scale at which pricing analysis is performed, but greater automation can also increase the need for oversight around how decisions are produced and applied.
That may be why insurers are not necessarily looking to remove human involvement from the process. More than half of respondents, 56%, said they favour a gradual approach that keeps human intervention in place for at least the next three years.
The findings also raise questions about how the actuarial role could change as AI takes on more of the technical work involved in pricing.
Rather than focusing primarily on producing a model, actuaries may increasingly need to consider what happens around it: how data enters the process, how outputs are validated, who has authority to act on them and how decisions can be explained.
That does not make the model less important. Instead, it changes where its value is realised.
A highly accurate model that remains stuck in a lengthy deployment process may have less commercial impact than a model whose outputs can be assessed, governed and put into use efficiently.
For insurers, the competitive question could therefore shift from how quickly a pricing model can be built to how quickly the organisation can safely act on what it produces.
That is a more complicated problem than improving model performance.
It involves technology, but also data infrastructure, internal processes, governance and the distribution of responsibility across different teams.
As AI becomes more embedded in insurance pricing, the distinction between modelling and decision-making is likely to become increasingly important. Technology can produce recommendations at speed, but insurers still need to determine whether those recommendations are appropriate, how they should be applied and who is accountable for the outcome.
For Earnix director Mathieu Edmond, this points to a change in where actuarial advantage will increasingly sit: not simply in producing a strong model, but in building the processes that allow insurers to deploy, govern and explain its recommendations effectively.
Copyright © 2026 FinTech Global
