Like many in our industry, BBH has spent considerable time exploring how AI and machine learning can enhance middle and back-office services. Here are three success factors that helped us achieve our goals.
Success factor 1: Invest in business problems, not AI
As we looked to implement AI, we learned that selecting the right use cases was crucial. Too often, firms focus on deploying an exciting technology without thoroughly vetting what it’s solving for.
One such use case was in our fund accounting business. We had nearly one million “miscellaneous” cash wires flowing into our accounting platform. Each one needed to be manually coded, researched, and resolved by team members. To solve this challenge, we developed the LINC (Language Identification Network Center). LINC uses natural language processing in a supervised machine learning framework to categorize cash breaks while making recommendations to operations specialists. The machine does this by recognizing words contained in wire transfer text, then the algorithm codes the description. The result: 95% of entries are automatically coded compared to 65% before we deployed the solution. The second phase moves beyond coding exceptions to now teaching the machine to resolve the breaks.
Success factor 2: One step at a time – operating through iteration
Building support for AI adoption can be a daunting task. The prospect of expense creep, risk of failure, and redesigns can often detract from the success of a project. With LINC, we added capabilities in stages and refined them incrementally. This allowed us to celebrate victories along the way while keeping our goals aligned with the big picture.
Success factor 3: Plan ahead for adoption
When it comes to enterprise adoption, we learned that nothing is as essential as formalizing processes for implementation. First, we partnered with our risk and audit control functions – areas of our firm that typically wouldn’t be involved so early – in establishing proper governance. While this added time at the outset, it also brought higher degrees of understanding of how the technology worked and how best to evolve the control environment. We also established a training program to introduce new roles that our AI created. This was no easy task; after all, we were asking many employees to adapt, acquire new skills, and adopt a “machine learning” mindset. Our analysts no longer need to know the steps to carry out a task, but instead how to analyze the performance of the system to ensure it’s accurate and learning correctly. Finally, we standardized metrics to monitor AI accuracy, application utilization, and efficiency. In this manner, we could follow in real-time that the tool was being used as intended. Looking to the future, these metrics are critical for the long-term success of our AI adoption, allowing us to know when and how often to retrain a model.
"Building support for AI adoption can be a daunting task"
Where we go next
With our first AI projects underway, we are building momentum for further AI adoption. This includes eliminating manual reconciliation activities across the enterprise, and leveraging solutions already made to attack like problems. For us, it’s essential to ask: Where can we improve continually? What can we do better? How do we continue to evolve? And most importantly, how can we create a better experience for our clients?
The views expressed are as of September 23, 2019, and are a general guide to the views of Brown Brothers Harriman (“BBH”). The opinions expressed are a reflection of BBH’s best judgment at the time. This material should not be construed as legal advice. Please consult with an attorney concerning your particular circumstances.