Xceptor: Automating Data Automation
How does a company that was at the forefront of data processing and automation adapt with the emergence of machine learning to further enhance its product offering?
How does a company that was at the forefront of data processing and automation adapt with the emergence of machine learning to further enhance its product offering? That is the challenge and opportunity that UK-based Xceptor is addressing now. Founded in 2003, Xceptor provides data automation solutions that streamline data intensive processes, removing human error and greatly improving operational efficiencies. The company primarily serves clients in the financial services industry, including top tier global financial institutions such as HSBC, J.P. Morgan, BNY Mellon and Deutsche Bank. For an example of its platform in action, Xceptor automated the trade confirmation process at ICBC Standard Bank, replacing costly, error-prone manual processes while also increasing oversight, audit and regulatory compliance capabilities [1].
The emergence of machine learning is particularly relevant to Xceptor’s product development process as it will enable the company to develop better, more robust solutions to address a broader range of problems for clients. As large financial institutions are under more pressure to innovate and remain relevant for customers, Xceptor’s clients will be pushing for more innovative solutions from their software providers. By improving capabilities and innovating to overlay additional services powered by machine learning in their product development, Xceptor will be well positioned to serve the continually evolving needs of their customers.
One key area where Xceptor is currently employing machine learning in their product development process is in enhancing its fraud detection capabilities. With the immense amount of data that Xceptor processes for financial institutions, it is well placed to utilize machine learning techniques to assist its clients in detecting fraudulent activity. While Xceptor’s software platform primarily utilized a rules-based approach, it is expanding to include native machine learning [2]. Tech industry resource Gartner has identified that strict rule-based data processing systems are most vulnerable to ongoing fraud attacks given their relative inflexibility, so by utilizing machine learning Xceptor will be able to develop more flexible products able to address constantly shifting fraudulent activity, enabling their clients to react more quickly [3].
Xceptor is also currently using machine learning in its product development to analyze intent as it processes certain types of data into its platform. By enabling the platform to assess intent, it can identify situations where human intervention is needed. To do so, the company is incorporating natural language processing (NLP)-based machine learning. This addition strengthens two of Xceptor’s core advantages. One, it improves Xceptor’s ability to handle diverse data streams and turn the data into high quality, reliable data that can be used for a wide range of purposes. Adding NLP to decipher intent in the data intake and cleansing processes adds another benefit for clients. Two, it improves usability for clients’ operational teams, a core benefit as the software platform does not require intensive support from clients’ IT teams. Adding NLP capabilities to decipher intent in incoming data sources makes the technology more user-friendly for the operational teams that use the Xceptor platform on a day-to-day basis, which is a core benefit of NLP [4]. Over the medium term, the company is also exploring how the emergence of blockchain will impact its business and service offering [5].
As the company looks at its future product development, I would recommend analyzing how other elements of artificial intelligence (AI) beyond machine learning could be incorporated. While Xceptor is looking at its options including blockchain, I would caution to be selective in its application. The core Xceptor product is exceptional at enabling companies to automate data processing, and many companies are still very much in need of this core automated processing capability. Instead of a dramatic overhaul in product development incorporating AI, it would be better for them to selectively and incrementally introduce aspects of AI. Driving incremental improvements that augment its current capabilities will do more to drive its next range of successful products and use cases rather than initiate transformational change centered on AI [6]. This preference for incremental improvement is shared with a majority of executives at the moment, as 51% of executives surveyed by Deloitte in 2017 indicated that the primary goal for AI is to “enhance the features, functions, and performance of our products” rather than dramatically overhaul them [7].
The company openly acknowledges that machine learning innovations are not needed in every situation, “lots of the processes financial institutions believe could be tackled do not need the latest, cutting-edge machine learning. This would be like using a sledgehammer to crack a nut.” However, the question remains of how does one balance the trade-off between its core rules-based approach against selectively incorporating machine learning and AI capabilities to enhance the product, without overcomplicating it and potentially detracting from user functionality?
Word count: 777
Sources:
[1] Xceptor. “Case Studies: Automating the Trade Confirmation Process.” https://www.xceptor.com/insights/case-studies/icbc-standard-bank/, accessed November 2018.
[2] Xceptor. “Using Machine Learning to Augment Automation.” https://www.xceptor.com/blog/archive/using-machine-learning-to-augment-automation/, accessed November 2018.
[3] Gartner. “Market Guide for Online Fraud Detection.” Published 31 January 2018. https://www.gartner.com/document/3849295?ref=solrResearch&refval=211640081&qid=417ad338a86c38b38d762cc56aba771e, accessed November 2018.
[4] Gartner. “Top 10 Strategic Technology Trends for 2019.” Published 15 October 2018. https://www.gartner.com/document/3891569?ref=solrResearch&refval=211644425&qid=ac5fc317f973c97519905c337e22fd97, accessed November 2018.
[5] Warensjo, Rob. “Xceptor Looks to Build on Niche Position.” Megabuyte. Published 14 June 2017. https://megabuyte.com/news/5940d6a6e4b0bea3ffd794d8/xceptor-looks-build-niche-position, accessed November 2018.
[6] Gartner. “Critical Capabilities for Data Science and Machine Learning Platforms.” Published 4 April 2018. https://www.gartner.com/document/3870311?ref=solrResearch&refval=211618257&qid=c3649fff26f3992f48625fc5a8230523, accessed November 2018.
[7] T. Davenport and R. Ronanki. “Artificial Intelligence for the Real World: Don’t Start with Moon Shots.” Harvard Business Review, Vol. 96 (January-February 2018).
Student comments on Xceptor: Automating Data Automation