CACEIS' Digital Innovation team is actively studying the many potential benefits that artificial intelligence can offer our industry in order to enhance the services we offer clients. The team has recently developed Pick-AI which can extract specific data from text-based PDF documents and will be initially used to process client fund invoices.
CACEIS processes some 40,000 invoices per annum from various legal, advisory and audit services for funds under depository. Such invoices can be entered directly through OLIS or sent by post or email. Our new AI technology will enable CACEIS to handle the ever-growing data volumes with greater speed, security and accuracy.
Operations staff will use Pick-AI as an integrated add-on that interfaces with the group’s back-office applications, starting with OLIS. It highlights key information in the PDF invoices it displays, and then lists these elements along with the value. Staff can easily make corrections by selecting a different part of the PDF, or by correcting the value in the list. New fields are easily added to extract other data points. Every time that zones are changed or corrections are made, Pick-AI will use machine learning to continuously improve its ability to extract the exact data required.
CACEIS’ IT division employed agile software development practices including requirement discovery and ongoing solutions improvement through collaborative, cross-functional team meetings with operations end-users. "We identified the most suitable technical solutions on the market. Pick-AI is composed of several AI algorithms that use the document shape, keywords, text position and value alignments for processing. It fulfils Operations’ business requirements in terms of performance, capacity and workflow integration," comments Sébastien Forest, CACEIS’ Data Scientist.
Invoice processing is the first practical use case for Pick-AI, and there are many other areas we have identified where operations staff members’ daily admin can be supported by this capable AI technology. "Once trained up, Pick-AI uses deep learning and natural language processing algorithms to identify and extract a wide range of key information from any document, and it improves each time it receives user input. We hope to implement other AI-based operations support technologies in 2024," confirms Abdo Berberry, Head of IT lines - 3D & Client/Business Development Support.
Co-creation and agile practices are essential elements in ensuring end-user satisfaction, and CACEIS' Digital Innovations division is ready to study and analyse any technology-based needs CACEIS teams or CACEIS clients may have.