The success of any enterprise – be it a project or a business – relies on its design. Defining key requirements, workflows, and data sources makes all the difference – and that holds true for IT systems as well. But when it comes to developing software solutions for financial services, this design phase can be notoriously complex.
To help solve this challenge, back in 2018 SnT partnered with financial service provider Clearstream and business analyst consulting firm escent to develop a new tool to slash error rates and cut production costs in financial IT development, all the while boosting service quality. “Nearly 50% of budget overruns in IT projects are caused by inadequate requirements that ripple through system design and deployment. SnT has brought together a strong team of scientists and engineers to tackle this critical problem,” said SnT’s Prof. Lionel Briand, head of the Software Verification and Validation (SVV) research group, and principal investigator of this research project.
The joint initiative set out to solve industry-wide problems in financial technology. One scenario the researchers explored involved how artificial intelligence (AI) and natural language processing (NLP) could support requirements analysis for IT projects in the finance sector. In these projects, requirements stipulate exactly what stakeholders wish to change, what the user needs and what the capabilities of the system need to be. However, no matter how experienced the team is, human error is inevitable. All too often, the outcome can be sub-optimal, resulting in disappointment and delays. This project aimed to make requirements analysis more efficient and effective by providing analysts with semi-automated assistance.
Elene Pitskhelauri, business analyst at Clearstream, said, “We were looking for an assistant that could increase the quality of the analysis and project efficiency.” In practice, their team were looking for a system that would verify that their IT requirements would be written correctly, and be based on solid foundations – as well as being internally coherent and correct. They recognised how AI techniques could address this challenge and, more specifically, understood how the technology could improve the efficiency of the projects and decrease their time-to-market.
On SnT’s side, the project saw the collaboration between a strong team of scientists. Led by Briand, these include Dr. Seung Yeob Shin, a research scientist; Angelo Rizzi, a research engineer; and Alvaro Veizaga, a Ph.D. student.
Together with Clearstream and escent, they created DRONA, a digital assistant that guides and semi-automates critical IT analysis. The tool provides semi-automated support for the definition of system requirements, their verification and the generation of documents. It also allows to check the requirements’ compliance against a set of predefined criteria. Relying on AI, as well as on automated validation and verification assistance, DRONA ensures the consistency and completeness of requirements to a significant degree. In other words, it ensures that the right functionalities will be developed and properly validated.
With it, “Cost and time-to-market are slashed, and system quality and customer satisfaction are boosted,” said Thomas Henin, Investment Fund Services IT business analyst lead at Clearstream. “This is because DRONA gives analysts the tailored support they need to build solid foundations for each project,” he continued.
Over the course of 2021, the team further refined DRONA to make it more robust, functional and user-friendly. Members of Clearstream’s team are currently being trained on how to use the software. With its successful proof of concept, DRONA could fill a gap in the finance market to make financial IT development more consistent, efficient and quicker.