AI is reshaping the retail insurance buying journey before customers ever reach an insurer. Customers are increasingly using AI tools to research coverage, compare providers, and narrow their options. Inside insurance organizations, AI agents are moving beyond assistance to perform complex tasks in areas such as underwriting and claims.
Yet many insurers have not changed how they organize around AI. They run dozens of pilots, often within individual businesses or functions, but lack an enterprise approach that turns those efforts into scalable capabilities. The reasons are familiar: uncertainty about where AI can deliver proven value today, questions about how quickly the technology will evolve, and short-term performance pressures that favor narrow, low-risk use cases over more transformative bets.
Scaling AI requires a different approach. Rather than allowing property and casualty, health, distribution, and other businesses and functions to pursue AI independently, insurers can organize their efforts around a common set of capabilities that can be adapted and applied wherever they create value.
Making that shift requires decisions that cut across traditional business boundaries: where to concentrate investment, what to build once for the enterprise, what tools individual businesses should control, and which legacy initiatives should be resized or stopped. For transformations to succeed, insurance CEOs need to take the lead in making those decisions and ensuring that limited, short-term initiatives do not crowd out the company-wide capabilities needed to scale AI.
Six Priority Areas for CEOs
Based on our project experience, insurance CEOs should focus on six priority areas. These span how customers find and interact with the insurer, how work is redesigned and supported by AI, and how the technology organization itself operates.
The priorities are deliberately defined across the enterprise rather than around individual lines of business or functions. That distinction matters. When insurers develop common capabilities horizontally, they can support them with shared infrastructure and reuse underlying technology and expertise throughout the entire organization. Business units retain responsibility for how those capabilities are applied and adapted, but they no longer need to develop the same underlying solutions independently.
When insurers develop common capabilities horizontally, they can support them with shared infrastructure and reuse underlying technology and expertise throughout the entire organization.
The priority areas reinforce one another, incorporate existing use cases rather than displace them, and support the insurer’s broader strategic priorities. Together, they can reduce complexity and duplication, improve coordination between business and IT, and help turn fragmented AI activity into value at scale.
AI Visibility and Access
For insurers serving consumers, AI is becoming a new gateway to the market. As customers increasingly use the technology to research and compare products, insurers need to prepare for a buying journey in which third-party AI may influence which providers customers consider. This means keeping the insurer visible and accessible as AI reshapes that journey.
The response unfolds in three stages. First, insurers need to be discoverable, ensuring that AI tools and agents can find reliable, current information about their products when customers research their options. Next, they need to be conversational, creating AI-enabled experiences that allow customers to continue the journey directly with the insurer. Finally, they need to become transactional, enabling AI agents to interact directly with their systems to obtain information, generate quotes, and eventually bind coverage. This agent-to-agent interaction is likely to emerge first in B2B markets but could ultimately extend to consumer insurance as well.
The stakes are significant. If AI increasingly determines which providers customers consider, insurers that are difficult for it to find or engage with risk being excluded from the buying journey before a customer ever reaches them.
Enterprise Conversational Layer
If they allow each line of business to build its own capability independently, insurers that apply AI in customer service can potentially re-create the fragmentation that AI should help eliminate. A better approach is to create a common conversational layer that spans businesses and channels, handles routine requests through self-service, and hands more complex cases seamlessly to human colleagues.
The potential value is already substantial. In our experience, call containment of 50% to 70% and service-to-sales improvements of 20% to 30% are achievable, depending on an insurer’s starting point, volumes, and processes .
With an enterprise conversational layer in place, call containment of 50% to 70% and service-to-sales improvements of 20% to 30% are achievable.
A shared capability also creates benefits that individual tools cannot. With appropriate data and governance, insurers can draw on interactions across the customer relationship to deliver an increasingly relevant and personalized experience. That can be particularly valuable in businesses with structurally higher churn, such as P&C and supplementary health, where a shared conversational layer can improve service and support retention while also lowering servicing costs.
AI-Enabled Process Redesign
To successfully scale AI capabilities, organizations need to redesign core processes rather than automating them piecemeal. Customer-facing AI addresses only one part of the opportunity. Insurers also need to rethink the processes behind those interactions—and not every process should be automated in the identical way.
Because activities such as underwriting and claims require risk judgment, coverage interpretation, and specialized expertise, they warrant tailored solutions and dedicated teams. Our experience suggests that AI-enabled redesign can generate efficiency gains of 30% to 60% in these areas, with even greater gains possible in selected processes. It can also improve effectiveness and contribute to a better combined ratio by supporting better underwriting decisions, lowering claims costs, increasing new business, and raising customer satisfaction.
The surrounding administrative and processing work, such as data extraction, can often be standardized and automated using shared capabilities rather than addressed function by function. Insurers should centralize the infrastructure and reusable components for common processing.
AI-Assisted Service
AI can make human service more effective, not simply less expensive. Even as customer-facing AI and redesigned processes absorb more routine work, some interactions will continue to require human judgment and empathy. A disputed claim, a serious health condition, or another complex situation may call for a person who can understand context, explain options, and adapt the response accordingly.
AI can make those human interactions more effective. Agent-assist tools can listen to conversations in real time, retrieve relevant information, suggest next steps, and reduce the administrative work surrounding the interaction. The human service representative remains responsible for the conversation while AI provides support in the background.
The opportunity can be substantial. In our experience, early implementations have produced efficiency gains of 20% to 50% and reductions in call transfers of as much as 90%. But the goal should not be efficiency alone. Insurers can reinvest some of the capacity they release in higher-quality human interactions—potentially improving customer satisfaction, reducing churn, and allowing service representatives to concentrate on the situations in which personal support matters most.
AI-Supported Sales
One priority goal of implementing AI is to give sales agents more time to sell and advise. As AI takes over more transactional work, human advice may become more important for complex or financially consequential insurance-buying decisions. The opportunity is not simply to automate the sales agent’s role but to remove the work that prevents agents from spending more time with customers.
Insurers can equip their sales agents with AI-powered digital counterparts that perform tasks that do not require a licensed professional. These digital twins can prepare quotes, follow up on paperwork, identify and cultivate leads, and move a potential sale through much of the process before human advice is required. Sales agents can then concentrate on complex products, important financial decisions, and customer relationships.
AI-powered digital counterparts can prepare quotes, follow up on paperwork, cultivate leads, and move a potential sale through much of the process before human advice is required.
In our experience, these capabilities can reduce the administrative burden on sales agents by 40% to 50%, creating substantial capacity that can be reinvested in sales activity. For insurers that have already developed customer-facing AI capabilities, much of the underlying technology can also be reused rather than built again—creating a second source of value from the same investment.
AI-Enabled Technology Delivery
The final priority sits within the insurer’s own technology organization—setting a clear ambition for how far AI should transform technology delivery. Many insurers are still modernizing legacy systems and working to improve collaboration between business and IT. AI creates an opportunity both to accelerate that transition and to increase the speed, capacity, and quality of technology delivery.
The opportunity can be viewed across three levels of maturity:
- AI-Assisted. AI supports developers and teams with tasks such as knowledge retrieval, code completion, and recommendations, while people remain firmly in control. The potential productivity gain is 10% to 20%.
- AI-Driven. Generative AI produces more substantial outputs—including code, other development artifacts, and data-driven forecasts—under human oversight. Productivity gains rise could potentially rise 25% to 40%.
- Fully Agentic. AI agents can execute increasingly large portions of the software development lifecycle within clearly defined goals, controls, and governance. The potential gains become much larger at this stage, with productivity increasing by a factor of three to five.
Greater automation can also strengthen operational stability through improvements in code quality, test coverage, and repeatability. But the appropriate maturity level will differ by team and activity. CEOs should ensure that the technology function sets a quantified ambition and target maturity level for each team rather than pursuing maximum automation everywhere.
A CEO Checklist for Scaling AI
The six priority areas define where CEOs need to focus. Creating value at scale requires a broader set of enterprise levers—covering strategy, investment, accountability, technology, talent, and partners. CEOs can use the following questions to assess whether those conditions are in place:
- Have we made AI a strategic priority? Set the AI ambition at the top and align it with the company’s broader strategic priorities. The CEO and executive team should agree on where AI can make the greatest contribution to the strategy—and ensure that investment follows those priorities.
- Are we concentrating resources where AI can create the most value? Focus initially on a limited number of high-value priorities rather than spreading resources across dozens of initiatives. Redesign the work end to end with an AI-first mindset, starting from the assumption that AI can perform the work unless there is a reason it cannot. Give each initiative a clear ambition, adequate resources, and an owner accountable for results.
- Are we funding the transformation by redirecting existing spending? Review the current project portfolio and resize initiatives that AI can deliver more efficiently or stop those that no longer support the company’s priorities. In our experience, 30% to 40% of resources planned in existing project portfolios can potentially be repurposed in this way. Use those resources to fund subsequent waves of investment, and prioritize AI initiatives that generate near-term efficiency gains.
- Have we put clear accountability and orchestration in place? Keep responsibility for implementation and results in the businesses, supported by a central capability that coordinates priorities and enables reuse across the enterprise. Integrate business, technology, and data roadmaps so that the organization is working toward the same priorities.
- Have we built the common technology and data foundation? Establish shared AI infrastructure, standard components, and reusable interfaces rather than allowing each business to build its own foundation. Make the data needed to create value discoverable, accessible, and of sufficient quality to support AI at scale.
- Are our people learning as quickly as the technology is advancing? Don’t wait for the technology to mature before building organizational capability. People’s mindsets and skills are harder to upgrade than technology. In BCG’s experience, roughly 10% of the effort in an AI transformation should go to algorithms, 20% to technology and data, and 70% to people and processes. That makes recruiting, reskilling, upskilling, cross-functional collaboration, and sustained change management central to the transformation. Just as important, employees need enough hands-on experience with AI to understand how their roles and ways of working need to change.
- Have we assembled the right ecosystem of partners? Decide which underlying technologies should be built internally and which should come from external partners. In many cases, external providers can supply the technology or infrastructure that allows the insurer to execute differently and faster. Select a limited set of complementary partners that can support the company across businesses, avoiding both dependence on a single provider and a proliferation of overlapping vendors that recreates silos at the enterprise level.
In the years ahead, what will increasingly set insurance CEOs apart is not whether they embrace AI, but how effectively they enable their organizations to apply it—faster, better, and at greater scale. Making the shift from isolated use cases to shared capabilities that can be reused across the enterprise will be a defining leadership challenge. Insurers that succeed in building and scaling those capabilities can turn AI from a collection of initiatives into a source of lasting competitive advantage.
