Generative AI: Value Unlock through a Cohesive E2E Ecosystem

In this era of increasing customer expectations, complex risk scenarios, and digital revolution, the insurance industry stands on the brink of a transformative shift. Generative AI, once a distant, abstract concept, is now at the core of this transformation, creating competitive opportunity across the industry, from underwriting to claims.

In our last article, we provided insight into how leading insurers are already seizing the moment and have started deploying big-win GenAI applications that scale well. This time, we focus on how these insurers are going beyond isolated use cases and driving true transformation through building end-to-end GenAI ecosystems. These GenAI ecosystems, anchored in value creation through prioritized use cases, are enabled by end-to-end people, governance, and technology components. Each of these components requires specific internal and external resources to bring them to fruition, all synchronized through centralized orchestration.

1. Value Creation through Prioritized Use Cases: Use case prioritization is typically a two-step process: evaluate the use cases through a value and feasibility lens and then identify and navigate major challenges (e.g., ethical and legal deal breakers, correctness criteria). The value and feasibility lens often yields three archetypes: A) Leverage off the shelf and become expert user (for example, code assistance, self-service chatbot B) Augment current offerings through custom builds (for example, assisted claims resolution) C) Commercialize into products and services for clients (for example, a reimagined product set) Many insurers are prioritizing archetypes A and B simultaneously, while setting themselves up to explore archetype C.
2. Talent, Organization, and Change Management: Insurers are managing the profound effect of GenAI on talent and organization, necessitating both changes in existing roles (for example, UX designers needing to have the ability to design AI interfaces) and the creation of new roles (for example, leadership roles such as chief AI officer, Large Language Model (LLM) engineer). In addition, they will also have to consider GenAI’s impact on talent acquisition, performance management, and personalized training. Furthermore, insurers are also equipping themselves for fit-for-purpose change management to effectively communicate and drive adoption.
3. Responsible AI: Insurers are quickly realizing that GenAI’s exponential value must be aligned with an organization’s purpose and values, realized througha) agreeing upon AI-specific ethical principles;b) setting up appropriate controls applied to the end-to-end product life cycle; andc) integrating responsive AI (RAI) into existing governance and risk management practices.With rapidly changing legislation through common regulatory themes such as explainability and interpretability, fairness, and avoiding unjust bias specifically impacting insurers, RAI is becoming critical to successfully harness the power of GenAI.
4. Data Governance: While data governance is already a key priority for insurers, it becomes even more critical for GenAI given the increasing unstructured, varied, and multimodal data inputs (for example, text AND writing, graphics, audio, etc.). This unstructured nature of data and the “black box” nature of GenAI introduces additional burden on data management and governance principles across data provenance, data classification, data lineage, data quality AND metadata completeness, and regulatory compliance.
5. Partnerships and Model Selection: Partnerships with hyperscalers and tech vendors are an essential part of the GenAI journey for insurers along with the LLM models (supported by the selected partners) serving the specific use cases. Insurers are using a comprehensive set of selection criteria (for example, geography, hosting, cloud portability) to choose partners and select models (for example, model output, model size, performance, fine tuning).
6. Platform, Infrastructure, and Integration Requirements: Insurers are rethinking their technology stack design to enable and deliver on GenAI use cases, driven by key questions around types of user, models, latency requirements, architectural guardrails, data formats, risk exposure, etc. The answer to these questions is introducing new layers into the technology stack (for example, AI Layer with AI model, AI guardrails, etc.) and impacting existing layers (for example, data layer).

The transformative value of GenAI is harnessed when each of these ecosystem components is synchronized through a central orchestrator.

We will be delving into each component through a specific Insurance lens in upcoming thought perspectives. In the meantime, please feel free to reach out to the authors with any questions.

Authors

Related Content