Private banking is moving towards a model centered on “advice”, where human expertise is amplified by AI. Customers are increasingly demanding: they want proposals aligned with their family situation, wealth projects and convictions (ESG, philanthropy, digital assets). Regulations (MiFID II and sustainability) nevertheless impose strict limits on recommendations. For this establishment, the challenge is to offer a highly personalized service while respecting these constraints. <\/p>\n
AI solutions already illustrate this. Personetics<\/strong> claims over 150 million monthly active customers and 1.2 billion insights produced every month: the platform analyzes transactional data streams and offers proactive messages on spending, saving or investing. BNP Paribas Wealth Management<\/strong> uses AI to analyze the press to identify events likely to influence the markets, assist managers in preparing multilingual reports and propose optimal allocations. AI does not replace the advisor; it enhances his or her ability to filter information and offer relevant advice. <\/p>\n Asset Feature Store<\/strong>: aggregates and structures client data: liquid and illiquid assets, cash flows, companies monitored, ESG level, currency exposure, time horizon. The store also includes eligibility metadata (risk profile, tax residence, MiFID status) to ensure suitability. <\/p>\n<\/li>\n Decision engines<\/strong>: propensity and uplift algorithms that estimate the impact of a recommendation (subscribing to an IA equity fund, arbitraging a real estate exposure). Models must be transparent and auditable. <\/p>\n<\/li>\n Content generation<\/strong>: a generative model (LLM) produces natural language analyses based on CIO notes, macroeconomic studies and product sheets. The use of RAG architecture ensures that quotes and references are verifiable. For example, a client interested in AI will receive a commentary on a thematic fund focusing on artificial intelligence, along with performance and market context. <\/p>\n<\/li>\n Human loop<\/strong>: the generated content is controlled by the banker, who adapts the message and decides whether it is appropriate. The system records actions to improve models (supervised reinforcement learning). <\/p>\n<\/li>\n Channel and timing management<\/strong>: AI determines the best time and channel (secure e-mail, customer area, appointment) to contact the customer, taking into account their schedule, time zone and regulatory suspension periods.<\/p>\n<\/li>\n<\/ol>\n Customized alerts<\/strong>: the system detects a customer’s over-exposure to the USD currency when his future expenses will be in euros. It offers three options: hedging via futures, subscription to the EUROD stablecoin, or diversification into a euro-denominated fund. <\/p>\n<\/li>\n Thematic briefs<\/strong>: based on the CIO’s watch and current events (AI, climate, health), the co-pilot produces thematic briefs for interested customers, citing sources (market notes, ESG reports). This enables customers to position themselves on emerging trends. <\/p>\n<\/li>\n Targeted campaigns<\/strong>: segmentation of customers by typology (entrepreneurs, endowments, cross-border wealth). AI identifies those who could benefit from an estate planning solution or tax residence transfer, and suggests personalized messages. <\/p>\n<\/li>\n<\/ul>\nComponents of a personalization platform<\/h3>\n
\n
Concrete applications<\/h3>\n
\n
Start-ups and tools to know<\/h3>\n