Artificial Intelligence: Transforming the Asset Management Industry

In 2016, scientists at SONY CSL Research Lab released “Daddy’s Car,” the first full song composed by artificial intelligence. The company’s AI software, called Flow Machines, created the song using machine learning. 

“They ‘taught’ it using a database of songs, including a lot of the Beatles,” said Kelley Conway, Managing Director, Technology Advisory Practice at Accenture, during a General Session at NICSA’s 2019 Strategic Leadership Forum.

Flow Machines does still require professional mixing and mastering via human participation, so the technology isn’t slated to replace musicians anytime soon. But “Daddy’s Car” underscores how rapid advances in computing power are enabling all new applications of technologies across many industries — including asset management.

“AI is not just one technology — it’s a number of technologies with focused applications,” Conway said. In terms of practical applications, Conway said AI is already transforming the way asset managers service customers, manage portfolios, and distribute products. 


Alan Anderson, Global Director of Enterprise Solutions at IPsoft, discussed his company’s experience developing the dynamic AI system Amelia, which helps deliver business value while creating new lines of revenue and enhancing the customer experience.  “With platforms like Amelia, customer service is paramount,” Anderson said. “Unless you do artificial intelligence really well, people are not going to use a chatbot more than once.”

To ensure Amelia meets customer expectations, IPsoft has programmed the software to perform tasks according to precise specifications, including cognitive learning abilities, autonomic task management, and emotional intelligence. “It has to be able to handle unpredictable things, and even change context — because in the middle of a conversation suddenly a spouse may walk by and say, ‘Hey, ask if we can we pay with American Express,’” he said. 

Ken Arakelian, Director of Cognitive Innovation Group Engagements, Nuance, said that as companies digitally transform, customer experience has become a major battleground. “Loyalty is going to come from how easy it is for your customer to work with your company,” he said.

One of the best examples of customer experience as a differentiator is Uber’s disruption of the taxi industry. “Same product, but the experience is completely different,” Arakelian said. “With a taxi, you stand out in the rain and raise your hand hoping somebody picks you up. In an Uber, it’s your ride, you’ve already decided where you want to go, and the payment is already worked out.” 

The story’s the same with asset management — the easier you can make the process for customers, the more loyal they will become. Robert Zembowicz, Chief Enterprise Architect at Vanguard, said machine learning gives us the ability to improve service beyond the limitations of human thought.

 “A couple of years ago at an AI conference, a leader from one of the major banks was talking about how they built a deep learning model to predict a customer’s next transaction,” Zembowicz said. “We all talk about collecting multichannel data — what if we use that data to build artificial intelligence solutions using machine learning that can capture much more complex relationships?”


Conway said of the more interesting aspects of AI in the recent past has been the introduction of sentiment analysis, or the process of computationally identifying customer attitudes. Arakelian said integrating sentiment analysis into AI-based asset management tools can prove challenging.

“Getting everyone to agree on the sentiment of a phrase is still tricky — and if you can’t get humans to agree on the sentiment of a phrase, then it’s hard to teach the machine to do so,” he said. “So the maturity we see today is a lot of good applications of sentiment analysis on post-interaction logs.”

For example, say you’re sorting through last month’s 10 million interactions and need to identify a rough scope of where you want to focus. “It will tell you, out of those 10 million, this 1 million look like they’re upset — and then you go look through that to find where you might have a retention issue,” Arakelian said. 

Zembowicz said sentiment is also about the client’s tolerance of market changes. “In the past, we would give customers a questionnaire to gauge their sentiment toward the market,” he said. “Imagine now, if we could use a machine solution to look up information such as previous transactions the person did, under what circumstances those transactions were made, and why they were trying to do those transactions to extend the notion of sentiment?”

Anderson said that at the end of the day, it’s important that new AI technologies are seamlessly integrated within existing systems. IPsoft’s 1Desk platform, for example, automates IT and business processes end-to-end by combining autonomic and cognitive technology into a unified platform.  

“That’s what we set out to create with 1Desk — to have the cognitive, the autonomic, the robotic process automation, the integration frameworks, everything, in one system,” he said. 

Note: Although the observations contained in this work represent the best thoughts of the individuals comprising the NICSA panel, they do not necessarily reflect the views of NICSA or any of its member organizations. Matters addressed in this work may touch upon legal or regulatory matters, however nothing herein is intended to be or should be construed as legal advice. You should contact your own counsel in order to obtain legal advice regarding these or any other matters.

Leave a Reply

NICSA: 8400 Westpark Drive, 2nd Floor McLean, VA 22102 • Tel: 508.485.1500 • Fax: 508.485.1560