“With a program of this scale, we can expect to see an increase of 10-15% on the bottom line. The use of generative AI technologies has been particularly important because it helped us with our users and sponsors, not only to explain things, but also for adoption.”
Adrien Vesteghem, Director of the AI Center of Excellence and AI Program for Efficiency at BNP Paribas

The BCEF (Banque Commerciale En France) is an entity of the BNP Paribas Group that is a leading global financial institution, offering a comprehensive range of banking, investment, and financial services, renowned for its strong international presence and commitment to sustainable and innovative banking solutions.

To launch its AI acceleration program, BNP Paribas turned to Artefact’s teams of experts to help them build an AI Factory. “The AI Factory is a response to our desire to move beyond the concept and PoC phases and to be able to generate value rapidly, securely and reliably by integrating with our information system,” says Anne-Sophie Bourdet, Data & AI Leader – AI Factory Director at BNP Paribas.

Challenge: Meet the objectives of the BNP Paribas’ AI acceleration program

Since its creation in 2000, with the merger of banks anchored in the European and global economy over two centuries, the BNP Paribas Group has been a global European leader, and an essential player in serving the economy, its customers and the world in which we live. BNP Paribas is fully invested in technology and innovation in order to continually optimize the experience of its customers and employees. It mobilizes the best of technology to offer them a seamless, customized, and secure experience.

According to Adrien Vesteghem, “There are four major reasons for launching this AI acceleration program:

  • The first is the value of AI. With a program of this scale, we can expect to see an increase of 10-15% on the bottom line.”

  • “Next are the benefits of open source AI solutions: security, efficiency, and most importantly, business value”.

  • “Then there’s the use of generative AI. This is particularly important because it helps us with our users and sponsors, not only to explain things, but also for adoption.”

  • “And finally, because BNP Paribas is already a few steps ahead of its competitors, this is the right time to consolidate, or even increase, our competitive advantage,” he concludes.

Solution: Mobilize the skills to build an AI Factory and industrialize viable GenAI use cases

One of BNP Paribas’ main challenges was to enrich its team with specialized competencies. “Our challenge is skills… data scientists, ML engineers, data engineers, new profiles that will strengthen our IT teams, which we’ll continue to integrate over the next few years to address our ambitions,” explains Anne-Sophie Bourdet.
She adds, “We need to be able to offer robust and secure environments to industrialize the solutions developed by our data scientists.
These are new requirements in terms of infrastructure (GPU) and foundation, to enable us to achieve the right level of end-to-end industrialization, while keeping our costs under control.”

Joffrey Martinez, Partner & Global Financial Services Lead at Artefact, notes: “To industrialize topics around GenAI, three major themes must be addressed:

  • Architecture optimization to control cost, customer experience and quality;

  • Value steering for the priorization of different use cases;

  • Organization to ensure sustainable diffusion of these transformation projects within the company.

The approach taken to manage the 70 or so new use cases within the bank was to use a roadshow to collect a large number of use cases from the bank’s various business units. Jérémie Cornet-Vuckovic, Data Consulting Director – Strategy and AI Projects at Artefact, describes the three-step process:

  • “Step one is ideation. We work with the businesses on the use case of tomorrow that could be highly valuable for them.”
  • “Step two is qualification, both in terms of technical feasibility with the AI Factory and in terms of the value created by each use case, whether in PNB, so additional revenue, or operational efficiency, or in terms of NPS. Each use case is thoroughly studied.”

  • “Step three is to go to the AI Factory with the AI Center of Excellence to define production roadmaps with major axes, whether it’s generative AI, document processing, client interactions, marketing, or fraud. Each axis is handled industrially between the businesses, the Factory, and AI Center of Excellence.”

Deployment of first Gen AI use cases with significant advances in fraud detection and improved customer experience

As for every major bank, a crucial challenge is phishing fraud, specifically the detection of fraudulent transfers made by customers on applications and websites. Jérôme Lhuillery, Data Science Lead – AI Center of Excellence at BNP Paribas, notes: This detection is now done in real time thanks to machine learning. It’s the first time we’ve implemented a machine learning mechanism in addition to our existing mechanisms, allowing us to achieve significant performance gains.
We’re also using a lot of GenAI to experiment with LLMs to provide more effective responses to clients and help our advisors work better every day.”

The teams at BNP Paribas appreciate Artefact for their dual expertise in data science and methodology, especially their competence in ML engineering, which optimizes and improves all pipelines and statistical engineering in AI and IT in order to deploy them in production. “For us, this is truly a win-win situation that allows us to meet very high technical and technological challenges,” Jérôme Lhuillery assures.

Adrien Vesteghem adds: “We chose Artefact for the quality of their offering, the clarity of their vision, and the depth of skills they provide to us in these AI acceleration programs.”

“The train’s running on the industrial and ethical track… The next step is to validate all these elements and we’ve already seen some pretty spectacular results with the use of Gen AI.”
Adrien Vesteghem, Director of the AI Center of Excellence and AI Program for Efficiency at BNP Paribas