IBM’s purchase of Promontory Financial Group put its machine learning engine Watson on the banking industry’s radar. The tech company’s purchase of the fintech firm Armanta earlier this month may help attract the largest American banks looking to use data aggregation and analytics to predict risk.
Eventually, Armanta and Watson will be applied to the same risk analytics solutions that IBM is using to help its clients dealing with financial risk.
“Armanta is adding to our financial risk solutions,” said Curt Burtmeister, director of research, innovations and financial engineering at IBM.
“Conceptually there is also a bigger initiative to bring Watson into our risk-based solutions. … You can see the two fitting very well because Armanta — whether it’s serving human masters or something like Watson — has the ability to ask what-if questions to hone decision-making."
Armanta, which will become part of IBM's Watson Financial Services portfolio, was founded in 2001. The company’s offerings now include a patent-pending shareable workspace called Armanta Sandbox which allows a user to change data, calculations or analytics and see the results in a test area. This gives the user the ability to evaluate different scenarios when predicting financial risk.
“We’ve been working very closely with IBM over the past few years, so this is not a complete sea change in the way we operate,” said Peter Chirlian, Armanta's CEO. “It’s just a natural evolution.”
As Armanta’s Sandbox continues to integrate with IBM, clients will have opportunities to interpret the data that Watson puts out, said Burtmeister.
For instance, if Watson’s natural language processing interprets a news article to be positive or negative, then clients can use IBM’s risk system to analyze how much the news might affect a particular stock in their portfolio. Armanta’s Sandbox can then analyze the results of the stress test and help clients understand how much the shift in that particular stock will affect the entire bank.
Another way in which Armanta and Watson may work together is by assessing credit risks. By looking for news articles (or other unstructured data) for issues like downgrades, defaults or bankruptcies in a company, Armanta can run an analysis using that scenario to see how the affected company may perform.
With 17 years to develop as a startup, Armanta is well established, said Ryan Gilbert, partner at Propel Venture Partners. Armanta is popular in the data science community, and Gilbert predicts that the acquisition will grab the attention of data scientists at the largest U.S. banks.
“The reality with AI and data science is that you need tools to implement the models,” Gilbert said. “Most data scientists know where the data is but they find it difficult from a practical perspective to bring the data together through aggregation and convert the data into something that’s easy to understand and is clean.”
Armanta’s competitors include companies that aggregate consumer data like Envestnet's Yodlee or Plaid — but few are driven by the institutions themselves.
“If IBM can lock in the top 100 institutions — those about $10 billion or more — they’ll be able to curate a good franchise,” Gilbert said.