Agentic AI could give CFOs more decision control

For decades, insurance decisioning processes – such as product changes, underwriting refinements, and new pricing strategies – were executed by IT specialists that could translate business requirements into code. 

In more mature organizations,analysts worked in conjunction with IT, using specialized business rule platforms. CFOs and policy leaders would set objectives, data science teams shaped product and pricing, and IT oversaw implementation.

These cumbersome processes involved a painful translation process between all entities and could take months.

According to an Earnix report, 25% of insurance teams collaborate regularly, with most working together on an ad hoc basis. This inevitably slows the decisioning process, by spreading every decision  across departments that already struggle to communicate – business, IT, and data science teams traditionally speak different "languages." 

These human and organizational factors account for about 70% of scaling challenges when implementing AI in insurance. Decision management models helped overcome siloing by expressing business logic on a unified platform, in a shared language that both business and IT teams could understand. But even with the decision models, agility and scaling remained a significant challenge. 

AI is now taking those strides a step further. By combining natural-language interfaces, simulations, and autonomous decision optimization, agentic AI lets business leaders achieve more control and agility, and reduce the cost of reacting to evolving market conditions. 

Overcoming silos

The history of insurance decisioning can be tracked across three stages:

1. Three silos: For any given decision, business, IT, and data science teams would approach the issue from their own perspectives with spreadsheets, models, and code that were unique to their department – and rarely matched. The process of translating one department's data into something comprehensible to the others was costly and unwieldy.

2. Two silos – decision management: Governed decision models became a unified reference, with shared rules and outcomes, effectively creating a lingua franca that both business and IT teams could understand. Decisions were still made manually as a collaborative effort between teams. But by removing the silos between business and IT, translation time shrank, errors were reduced, and control and agility increased.

3. No silos – agentic AI: Agentic AI now sits on top of decision management tools. Semi-autonomous or autonomous agents can continually assess and translate the operational data within a decision management tool and use it to optimize every decision. The three wheels (business, IT, and data science) effectively merge under an agentic enterprise umbrella, with the grunt work of assessment, translation, and action recommendations handled autonomously.
PwC reports 54% of insurance executives say AI is the technology most likely to reshape the industry over the next three years. Some adopters are reporting 14% higher retention and 48% higher net promoter scores (NPS) compared with non-users.

Bridging strategy and execution

Agentic AI–powered simulations are the bridge between strategy and execution. 

Instead of guessing how new discounts or underwriting criteria might affect profitability, business leaders are presented with simulated change impacts across thousands of customer profiles, or they can test new products without disrupting operations. This helps instantly gauge financial benefits before roll-out.

A CFO, for example, could ask an AI agent: "What would happen if we tightened underwriting criteria for coastal properties by 15%?" They'd instantly see simulations of premiums, loss ratio, and profitability impact, without waiting for IT to translate the code, or for data scientists to parse the complex dashboards.

Redefining collaboration

Decision management gave siloed teams a shared model. Agentic AI turns that static artifact into an active collaborator. 

Here's the breakdown:

• Business and finance leaders express goals and constraints in plain language.
  

• Agentic AI translates those written intentions and guidelines into analytical components, code and parsed data that IT and data science teams can understand.
  
• With the written business objectives now translated into code, IT teams can accurately integrate governance and security guidelines into the actual functional systems.

Rather than reduce the importance of IT or data science teams, it allows them to focus on high-value work, instead of just converting ambiguous requirements into code.

Democratized decisioning only works if management tools remain transparent and auditable, so that regulators can track compliance, and CFOs can explain the changes they implement. 

Insuring the future

Gen AI could add up to $4.4 trillion in value to the global economy annually, according to McKinsey. Insurers are now in a position to turn that potential into faster, more agile operations, where strategy and execution are tightly linked. Those that invest now to add agentic AI to their decision management strategy will come to define the next era of competitive advantage in insurance.

Instead of being dependent on a handful of technical experts to interpret ambiguous data, organizations can embed that expertise in the decisioning platform itself. Anyone with business acumen and an understanding of objectives can then confidently participate in shaping decision logic, without outsourcing to IT or data science teams, and losing agility.

The path forward is incremental: start with high-impact use cases, like pricing adjustments or new product launches, and measure success using business metrics, such as time-to-market and portfolio performance. Agentic AI will define the next era of competitive advantage in insurance. It will be interesting to see which insurers lead this charge and which ones sit on the sidelines.