In an era where digital transformation is redefining financial services, BNP Paribas is making significant strides in applying artificial intelligence (AI) to streamline operations, enhance customer interactions, and optimise risk management. The bank has actively been integrating AI-driven solutions into its core functions, aiming to create a more seamless, efficient, and personalised banking experience.

We were invited to a press conference with presentations by:

• Sophie Heller, Chief Operating Officer, BNP Paribas Commercial, Personal Banking and Services;
• Sébastien Perrigault, Director Consumer Finance, Banque Commerciale;
• Bertrand Cizeau, Director, Hello bank! France;
• Anthony Belpaire, Head of AI, BNP Paribas Fortis.

A Strategic AI Roadmap

BNP Paribas' AI initiatives focus on three primary objectives, and I like the fact that this is not considered a 100% efficiency game but that they are looking at this transformation by taking more parameters in the equation:

• End-to-end process optimisation – Automating and refining internal workflows.
• Ultra-personalised customer experience – Leveraging AI for tailored financial interactions.
• Operational efficiency – Enhancing business activities, fraud prevention, and risk assessment.

Efficiency is obviously an important KPI; by optimising processes, they can reduce errors, leading to better decisions and better customer experience, and the extra use cases on personalisation put the customer at the centre of developments, as it should be.

The journey of AI adoption within BNP Paribas has been ongoing for years, with a shift from traditional models to AI-driven solutions since 2016. This transition has enabled the bank to refine key areas such as marketing, risk assessment, and customer service.

Revolutionising Marketing and Risk Management with AI

AI has played a pivotal role in reshaping BNP Paribas' marketing strategies. Previously, the bank relied on basic product scoring models to determine customer interest in various offerings. Today, advanced AI models combine multiple data points—including real-time customer interactions, economic indicators, and network availability—to suggest the most relevant actions for each client at the optimal time.

Similarly, AI-driven credit risk assessment has transformed decision-making for small business loans. By analysing transaction flows on business accounts in real-time, AI provides a highly accurate evaluation of repayment capacity, reducing risks and expediting the loan approval process.

The Rise of Generative AI in Banking and the Challenge of Balancing Customer Experience and Compliance and Risk

The next phase of AI implementation at BNP Paribas involves generative AI, which can read, listen, and write. This technology is particularly useful for automating document analysis, email processing, and customer service interactions. To ensure security and compliance, BNP Paribas has partnered with AI firms such as Mistral to develop proprietary models that operate on the bank’s infrastructure, ensuring sensitive customer data remains within controlled environments.

Given the sensitive nature of banking data, BNP Paribas has implemented stringent security measures in its AI deployments. The bank’s AI systems operate within secured environments, using proprietary models trained on internal data. Additionally, a human-in-the-loop approach is maintained to verify AI-generated content, reducing the risks associated with AI "hallucinations"—incorrect or misleading outputs.

AI-powered virtual assistants are at the forefront of BNP Paribas' customer service evolution. The bank’s generative AI systems facilitate hyper-personalised 24/7 assistance, handling both informational and transactional queries. A notable example is Hello bank!’s AI chatbot, "HéloïZ," which has significantly improved with the introduction of generative AI.

Unlike its predecessor, which relied on predefined answers, the new version provides empathetic and contextualised responses, enhancing customer satisfaction. The improved AI assistant also demonstrates a deeper understanding of customer inquiries, eliminating the need for redundant clarifications. This results in faster, more natural interactions and significantly reduced support agent workload while improving customer satisfaction.

By next month HeloïZ will be rolled out to 10,000 clients, to receive feedback for improvements. That number will grow in tranches of 200,000 clients to reach 1,000,000 clients by the end of the year.

AI-Driven Mortgage Processing and AI Assistants for Employees

BNP Paribas is also leveraging AI to accelerate mortgage application processes. Currently, the time between mortgage approval and electronic offer issuance is 10 days, but with AI automation, the bank aims to reduce this to six days by mid-2025 and further to three days for straightforward applications by early 2026.

To achieve this, BNP Paribas has developed an in-house document recognition AI system, which automates the verification of key mortgage documents such as salary slips, tax returns, and purchase agreements. By using Optical Character Recognition (OCR) and machine learning models, this system ensures faster processing, fewer errors, and improved compliance with risk assessment standards.

Beyond customer interactions, BNP Paribas has rolled out AI-driven assistants for internal use, particularly within BNP Paribas Fortis, for 8,000 colleagues to be precise (out of a total of 10,000 for now).

The initiative aims to enhance productivity across various departments, including investment communication and software development. The AI assistant, "Yara Invest," is currently deployed among private banking teams to help synthesise investment reports, translate documents, and generate tailored financial insights.

One key advantage of Yara Invest is its ability to customise content based on individual client profiles, ensuring that investment information is presented that aligns with each client’s expertise level and preferences and to do this at a much faster pace.

Future Prospects: Towards Transactional AI

While BNP Paribas’ AI chatbots currently handle inquiries, the bank’s long-term goal is to integrate AI into transactional operations. Future iterations of AI assistants will be able to process customer transactions, such as fund transfers and payment authorisations, providing a truly autonomous banking experience.

To achieve this, BNP Paribas is working on a unified AI model across its various divisions, ensuring a scalable and cohesive AI strategy. The bank is also exploring ways to optimise AI processing costs, improve response times, and enhance data security.

BNP Paribas' extensive AI strategy reflects the broader transformation occurring in the financial sector. By harnessing the power of AI across marketing, risk assessment, customer service, and internal operations, the bank is setting new benchmarks for efficiency and customer engagement. With a strong emphasis on security and compliance, BNP Paribas is positioning itself as a leader in AI-driven banking, paving the way for more intelligent, responsive, and personalised financial services.

Join us at The Banking Scene Conference Brussels on May 22 for more practical applications, case studies and insights from industry leaders pushing the boundaries of AI in Banking.