- BNP Paribas’s artificial intelligence lab is developing foundational capabilities to reorient investment and corporate banking around customer information
- It is building translation, search and chatbots with a strategy designed to create value as other aspects of business become commoditized
A young team of artificial-intelligence experts is setting the stage for BNP Paribas to expand its business strategy for corporate and investment banking (CIB) beyond transactions.
The CIB division’s AI Lab, established just a year ago, is tasked with improving client experiences and internal efficiencies, says Edouard d’Archimbaud, head of the artificial intelligence lab in Paris. But the team sees a new direction in how the bank will add value.
“Today a bank generates revenues by facilitating financial transactions, but increasingly our business will involve helping clients extract value from information,” d’Archimbaud told DigFin. “This is a big shift.”
To that end, the BNP Paribas AI Lab is engaged in foundational work – the sort of thing that sounds more like a Google project than a bank’s. In a way, it is recreating what a Google or a Baidu does, but specifically to meet a bank’s needs.
One of the team’s first projects is to build a translation engine, which it has done in partnership with several businesses in the bank. The immediate use cases are to automate document retrieval, review contracts, and assist compliance. “A 150-page contract can be reviewed in 15 seconds,” d’Archimbaud said.
The AI Lab has been training its computers how to translate languages by using transcripts from European Union parliament speeches – all of which must be translated into all EU member languages – and with customer documents (with the client’s permission).
This is like what Google Translate does, but d’Archimbaud says BNP’s version is better adapted to financial services. “We are creating state-of-the-art algorithms for financial vocabulary,” he said. And once a translation program is developed, it becomes easier to scale it with new languages, provided there is sufficient data; the lab is now teaching it Chinese and Japanese.
From translation, the AI team is extending its work to enable a computer to recognize names of people or companies, as well as identify locations and other data from reading documents. This is leading it to develop a search function (as well as index internal data).
A third foundational capability is interfaces, specifically chatbots, which can allow a BNP banker or a client to use search to access relevant information in any language. D’Archimbaud says chatbot technology is years away from maturity, but his team is investing a lot of time in voice recognition.
“Ten years ago, people accessed information by scrolling the web,” he said. “Then activity moved to searching on your mobile phone. Now it is moving to conversational encounters.”
The goal is to have a user experience akin to asking Apple’s Siri or Amazon’s Alexa, only a banker or a client will be seeking contractual or transactional data, rather than reserving seats at a restaurant or buying movie tickets.
BNP Paribas hopes that by establishing these building-block functions, it will provide clients with a simpler, faster and more efficient means of engaging with the bank.
How to add value
From there, says d’Archimbaud, BNP Paribas hopes to use this more intense engagement to anticipate customer needs. This could also lead to new businesses, such as custody and analysis of client information.
Clients include corporations and financial institutions, such as institutional investors and hedge funds. As core financial transactions become more commoditized, this kind of service could become an important new source of revenue.
“Banks will want to provide a secure environment for clients’ data, and for all of our AI-related activities,” d’Archimbaud said.
For now, BNP’s AI Lab is working with clients and internal business heads to imagine, and begin to design, these services.
To realize this vision, the bank needs to generate more actual projects. More actual use cases would help scale projects. More work is also a good way to attract more talent.
“We need to build a startup environment within the bank,” d’Archimbaud said.