InsurTechTalk: Build versus buy in the AI era – strategic tech decisions for life and annuity carriers

Views on news
Manulife has launched a proprietary enterprise AI platform, marking a shift from building individual AI tools to running AI as a shared capability across its global operations. The platform gives teams a common environment to build, deploy and manage AI applications. Tools already reviewed for security, risk and governance can be reused in other markets without repeating the full approval process. Manulife’s move illustrates the broader shift from point solutions to integrated platforms, as well as the level of competitiveness and value creation insurers must demonstrate to stay in the race. It also highlights the pragmatism needed to deploy AI effectively, starting with the low-hanging fruit that can deliver significant value. While the resources required to build solutions in-house remain substantial today, doing so could become a more viable option over the next two to five years. In considering an in-house build, however, insurers must candidly assess the availability of talent, resources and the right architecture.

The changing definition of what a strategic partner is
Only a small fraction of carriers currently has the capacity to build their own platforms. Although AI can take care of much of the development work, businesses must also understand their requirements for managing complex core systems. Once built, these systems need to be maintained, upgraded and enhanced, drawing on new data and adapting to changing business needs. There is a reason that, even before AI, vendors developed new products through shared investment from groups of carriers. While prototyping and experimenting with technology have become easier, and capabilities such as intelligent document processing can now be implemented without a partner, other aspects remain too complex and expensive to manage in-house. If you cannot reliably answer the questions “What does good look like?” and “What makes it difficult?”, start by defining the type and level of external support you need – whether that means buying software, developing a solution or co-developing it. When choosing partners, it is also important to consider whether their business model is sustainable over the longer term.

That said, AI is also set to transform implementation over the next couple of years, changing expectations of partners. A partner becomes strategic when moving away from them would disrupt your business. The question, then, is not which frontier-lab partner you choose, but what your core systems are and who is best placed to support them. As priorities shift, what was once considered strategic may become ancillary. Conversely, for carriers with legacy technology stacks, partners can become strategic almost by default, as delays or changing priorities make it harder to move away from them. A good example of how strategic partnerships can be disrupted is DXC’s reported sale of its insurance software and services division to Eldridge.

For non-core solutions, AI can even double efficiency. Capabilities such as intelligent document processing, document summarisation and computer vision can act as force multipliers and are increasingly available as out-of-the-box features of large language models. When considering an in-house build, start by identifying the business’s most impactful priorities – whether that means reducing unit costs, scaling a particular area of the business or improving the adviser experience. Then compare cost, speed to market, ease of integration and return on investment to decide whether to develop or buy. It is also essential to establish a framework for custom-built software, including approved technology stacks and security standards. But don’t be misled by how easy it is to create a prototype: that is only one part of the job and does not address the more challenging aspects of implementation, maintenance and integration, where a partner may be essential.

The panel’s advice

• Organisations must assess their “fitness level” before making a buy-versus-build decision.
• It may be hard to get a clear view of vendors’ operations from the outside. Talk to them to see where they are on their AI journey.
• Find vendors to support workflows that you either cannot manage yourself or cannot deliver commercially viably. When looking for strategic partners, consider whether they are the kind of organisation you will be able to solve future problems with.
• Rely on data but also “search your feelings” when deciding who would make a great strategic partner.
• Equip your people with AI skills and give them the freedom to experiment and identify new use cases.