Jean-Jacques Pluchart
The spectacular advances in AI – and in particular in generative AI since 2022 – are disrupting the strategies, organisational structures and practices of an increasing number of industries, particularly in the banking and insurance sectors. The scale and speed of these transformations can be seen in the fluctuations in the margins, earnings and share prices of listed institutions. The most erratic fluctuations in certain stock prices reflect the uncertainty felt by savers and investors regarding the ability of banks and insurers to adapt their value creation chains and rebuild their business models.
Banking businesses are built on the secure management of personal data and the coverage of risks of various kinds. Traditionally, banking businesses are divided into retail banking and wholesale banking. However, they are becoming increasingly differentiated according to the bank’s predominant strategy, which may focus on volume or on the differentiation of its products and services. In the former case, they cover ‘document-intensive’ functions, and in the latter, ‘high-responsibility’ functions. The former encompass the administration of day-to-day operations, the generation of contracts, customer relationship management (CRM), accounting and financial analyses, etc. The latter involve trade-offs between transactions, the issuance of credit, legal, tax and financial arrangements, and strategic decisions, etc. The former can increasingly be replaced by automated processes. The latter can only be supported by dedicated AI-based models for recognition, classification, simulation, projection, correlation, etc. Distinguishing
between these two types of activities is becoming increasingly difficult due to the rapid progress of AI and LLMs, which are based on the massification of data, the acceleration of data processing, the proliferation of specialised AI agents and, above all, the ability to quickly code new programs using natural language (machine learning or automatic encoding). As a result, new functions are being ‘augmented’ by AI: the development of more sophisticated chatbots for interacting with prospects and customers, the security of data and data processing, the systematisation of securities rating, the automation of compliance (due diligence), the optimisation of securities settlement and delivery, etc., as well as the enhancement of the reliability of forecasting models (predictive trading) and the simulation of credit and market risks. Functions that were previously performed by specialists with rare skills are thus becoming ‘commodities’ provided by standard applications (benchmarks).
Advances in AI and LLMs are leading to the disintermediation of value creation chains, the reconfiguration of banks’ business models, and the reshaping of their ecosystem. These shifts can be observed in the changes in the margins and stock market prices of banking institutions and their subcontractors. SaaS software outsourcing licences are gradually being replaced by proprietary models generated through ‘Vibe Coding’ at low marginal cost. This transition to token-based pricing has already led to a drop in the MSCI USA Software index. For example, the share prices of Salesforce, Thomson Reuters and LegalZoom have been affected. The downward trend in margins and valuation multiples is beginning to affect software publishers, property and personal insurance companies, and financial and non-financial rating agencies. Insurify’s launch of a purchasing agent capable of instantly comparing millions of policies caused a sharp drop in the value of Willis Towers Watson and Aon. In the fields of accounting (auditors, analysts) and credit rating, the same phenomenon has affected certain agencies, such as S&P Global, Moody’s and FactSet. Retail banks, which focus on providing advice and credit to individuals and SMEs, are directly exposed to a loss of competitive advantage unless they demonstrate their ability to adapt quickly to the changes brought about by AI.
In contrast, investment or merchant banks benefit from barriers to entry based on the personalisation of client relationships (i.e., on trust and personalised historical data), on financial, legal and tax structuring (M&As, major projects, etc.), particularly at the international level, on wealth management, on the securitisation of receivables, on the management of derivatives, on strategic decisions, and even on certain functions related to shadow banking (management of investment funds or tax avoidance schemes, etc.). The quality of the banking relationship creates value when it is developed during a monetary and financial crisis or simply in a volatile market environment. The involvement of a human adviser provides the client with ‘mental and emotional well-being’ and greater confidence in the future.
This transformation of banking models prompts us to revisit the lessons taught by Michael Porter since the 1980s, which distinguish between corporate strategies based on volume and those based on service differentiation. It appears that, in the wake of advances in AI, these lessons are once again becoming increasingly relevant. Banks are being compelled to adopt strategies that focus on innovative and phygital activities, combining these two approaches.