In February 2018, Commonwealth Bank and ING, in partnership with fintech Ascent Technologies and law firm Pinsent Masons, completed a pilot that demonstrated the potential of Artificial Intelligence (AI) to solve an industry problem – how can organisations simplify the process of understanding and complying with their regulatory obligations?
The Financial Conduct Authority (FCA) was a key member of this experiment. It participated as an observer of the project. The FCA understands the benefits of collaboration in areas where an industry solution is beneficial for the overall system, such as compliance. Observing the project gave the FCA greater understanding of the potential applications of Natural Language Processing (NLP) and AI in making compliance more effective and efficient.
Complex industry regulation
RegTech is a focus of our Innovation Lab in London. Mukund Umalkar, Innovation Manager at the Lab, says: “We are looking at using emerging technologies to make compliance with regulations more effective and efficient. It is a common industry challenge and one that offers opportunities for the firms, regulators and start-ups to collaborate and solve together.”
In September 2017 we started talking with ING and realised the two teams shared a similar philosophy of collaboration and openness. We saw the partnership as a chance to cross-pollinate ideas, as well as to share skills, insights and costs to test innovative RegTech solutions in the market.
MiFID II test case
Commonwealth Bank and ING honed in on Ascent Technologies because of the way it leveraged AI, particularly NLP, and its clear understanding of the problem space.
Because banks work closely with law firms to interpret regulation to ensure compliance, we also invited Pinsent Masons to join the pilot.
The pilot kicked off in October 2017 in the controlled environment of the Lab. The objective was to test Ascent Technologies’ platform for part of Markets in Financial Instruments Directive II (MiFID II).
Understanding regulation and implementing the necessary policies, procedures and system changes is currently a very manual and time-consuming process. Compliance teams spend a lot of time performing administration tasks that go with understanding voluminous regulatory packages.
It also involves considerable liaison with lawyers. For example, it took our Compliance team several months to develop a Traceability Matrix by compiling a list of applicable obligations and tasks for MiFID II, not to mention the effort involved every time there is a change in the regulation.
In contrast, the pilot only took a matter of weeks to interpret and convert 1.5 million paragraphs of regulation into bitesize, actionable tasks for implementation. These are automatically updated every time there is a change in the regulation.
“The technology isn’t 100% accurate yet but it’s far better than starting with a blank spreadsheet,” says Mukund. “It is a massive productivity gain.”
Model for collaboration
The successful pilot demonstrates the tremendous potential for AI and NLP to make regulatory compliance far more efficient. It could free Compliance teams from administrative and repetitive tasks so they can focus on more complex, value-add work.
Equally valuable is the development of an effective model for collaboration between industry, regulators and new technologies.
“Instead of working on the same problems in silos, this model allows two or three banks to collaborate on solving shared problems,” says Mukund. “If we all do that, we don’t have to solve all the problems by ourselves, which is what banks are doing today.”