Jul 262016
Curt Hall

Cognitive computing is starting to impact the enterprise by changing the way data is analyzed and the manner in which employees and customers interact with computerized systems. This is happening across various industries, ranging from healthcare and retail to banking and financial services. Since I have been delving into the financial area of late, I wanted to provide a glimpse into how banks and other financial institutions are utilizing cognitive applications and commercial solutions within their organizations.

Key Drivers

Key trends driving cognitive into banking and finance include:

  • Massive data sets. Finance and banking firms are facing an increasing need to analyze very large data sets consisting of structured and unstructured data for exploration and discovery, market research, risk analysis, compliance, operational engagement, and customer service and customer experience management.

  • Growing number of commercial cognitive offerings. These include cognitive development platforms offered in the form of API-based cloud services available from such providers as IBM (Watson),Microsoft (Cognitive Services)CognitiveScale (Cognitive Cloud), and Expert System (Cogito). Cloud-based development services make it easier for developers to build cognitive applications and commercial solutions.

  • Industry-specific/focused cognitive products. Cognitive, service-based solutions targeted at specific banking and financial markets and applications are appearing with increasing rapidity and include offerings from 50wise, an IBM Watson partner; Expert System (Cogito Risk Watcher)Domo Semo Sancus (SafetyNet)iQventures (speechiQ)Kasisto (KAI platform); and Kenshoo. This is an accelerating trend, making it easier for firms to implement cognitive capabilities into their banking and other financial processes and applications.

  • Coming-of-age natural language processing (NLP) and machine learning (ML). There’s a growing appreciation of the benefits offered by NLP, ML, and interactive natural language UIs, leading to their use in adding self-service capabilities to decision-support and advisory applications. These capabilities are proving beneficial to both employees and customers.

  • Customer experience and engagement. Combining cognitive with mobile, speech, and touch technologies supports the implementation of smart, multi-modal interfaces designed to deliver an enhanced customer experience to mobile, Web, and other banking and financial applications.

Primary Applications

Cognitive systems can analyze extremely large volumes of structured and unstructured data, including research reports, product information, industry and government regulations, and customer profiles — paving the way to the discovery of key trends and findings as well as the identification of connections between customer needs and available financial services. Enterprises can use cognitive systems to support internal operations and employees involved in assisting customers as well as consumer-facing processes and applications — in particular, mobile apps and other self-service offerings. Key domains for applying cognitive solutions in banking and finance include:

  1. Research and discovery
  2. BI, advisory, and other decision support
  3. Risk assessment, compliance, and fraud prevention
  4. Customer service and customer experience management

In this article, I explore the third of these domains.

Risk Assessment, Compliance, and Fraud Prevention

Risk assessment, compliance, and fraud prevention are leading areas for applying cognitive technologies in banking and finance. This makes sense because the cost of noncompliance can result not only in fines and other penalties, but also in damage to a firm’s reputation, suspension, missed business opportunities, and, in extreme cases, criminal prosecution. Assessing risk is also difficult because evaluating the risk profile of a party or company seeking loans, banking, or other financial assistance goes beyond merely analyzing an individual’s or company’s financial documents; it is equally necessary to analyze the profiles of a company’s officers to uncover factors indicative of potential risk. In short, risk assessment requires analyzing a large number and manner of financial, regulatory, investment, and other data. Moreover, it demands knowledge of the rules and regulations pertaining to risk assessment and regulatory practices, as well as how to apply them. Several providers offer commercial cognitive-based solutions designed for assessing risk in financial and banking operations.


Domus Semo Sancus, a FinTech company, has developed what it refers to as a “know your customer compliance and risk management engine” that draws on a large database to assess risk. SafetyNet integrates multiple IBM Watson language APIs to help organizations quantify, with a level of confidence, the risk that a person or entity of interest is engaging in illegal practices. The government of Turks & Caicos Islands uses SafetyNet to improve operations by vetting potential investors. By harnessing Watson’s NLP and ML algorithms, SafetyNet ingests and analyzes extremely high volumes of information at a rate of thousands of pages per second, enabling it to expose patterns, connections, and insights around persons and entities of interest at a depth and speed no human analyst can possibly achieve.


50wise, another FinTech specialty firm, has developed a commercial cloud-based offering to provide query support to card payments professionals. Built using IBM Watson’s Question and Answer (Q&A)API,[1] the 50wise service is designed to answer questions that a bank, financial institution, or payments processor may need to ask about the rules and regulations surrounding the acceptance and processing of non-cash payments. Indra Financial Services, Madrid, has licensed this service to provide its employees with real-time answers and expertise to customer inquiries in order to improve bank card payments processes.

To implement 50wise, developers first taught Watson to understand the language of the payments industry by having it ingest thousands of pages of relevant rules and regulations, including those issued by card schemes and industry regulators (e.g., PCI Security Standards Council). This training created a massive knowledgebase that users can access simply by asking questions in natural English language sentences. Watson returns a set of suggested answers with a level of certainty measure and the source(s) from which the answers were acquired to enable analysts to conduct further investigation.

50wise also offers an API that financial companies can use to integrate payment industry expertise into their operational systems and decision-support applications. This enables companies to enhance their customer service applications, dispute/chargeback processing systems, and product development processes with the latest expertise about the rules and regulations governing the acceptance and processing of non-cash payments.

Cogito Risk Watcher

Expert System’s Cogito Risk Watcher is a commercial solution for enterprise operational risk and compl­iance analysts. It supports deep and wide analysis by applying semantic understanding and reasoning to thousands of sources and millions of documents and then narrowing them down to the ones that matter. An intuitive GUI supports interactive exploration with maps and multiple views of findings. Risk Watcher’s ability to analyze external and internal structured and unstructured information is intended to facilitate building better risk profiles of current and potential customers, vendors, and other stakeholders by uncovering facts, connections, hidden relationships, or other relevant information between different sources and providing multiple scenarios as data emerges for risk mitigation or threat discovery.


[1] Last summer, IBM introduced four new services that replaced the Watson Q&A service. The new services can be trained on your industry- or application-specific data and do not require deep knowledge in ML and linguistic models. Consequently, the Watson Q&A service tile was removed from the Bluemix catalog in November 2015.

[For more from the author on this topic, see “Cognitive Technologies in Banking and Finance, Part I.”]


Curt Hall

Curt Hall is a Senior Consultant with Cutter Consortium's Data Insight & Social BI and Business & Enterprise Architecture practices. His expertise includes BI, data warehousing, data mining, and other analytical technologies and products.


  One Response to “Cognitive-Based Solutions for Assessing Risk in Banking and Finance”

  1. What if the data of interest is the computer code?
    How do we find the ‘bugs’ or faults in an enterprise/s code that lead to interruptions, seemingly accurate output, and vulnerabilities that cyberattacks can exploit or ?

 Leave a Reply

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>