AI-based solution for analysing digital documents

Xceptor delivers no-code data automation software for the banking, financial and insurance industries – simplifying and speeding up complex tasks.

Professors Baihua Li and Eran Edirisinghe collaborated with Xceptor on an 18-month KTP to develop an AI-based solution that automatically analyses and extracts information from digital documents.

The KTP - conceived by Professor Eran Edirisinghe (now at Keele University) and Dan Reid (Xceptor founder and CTO) - has successfully delivered new technical capabilities while enhancing our fundamental understanding of image recognition processing and artificially intelligent (AI) deep neural networks.

Benefits for the company

Keen to consolidate its position as a market leader, Xceptor wanted to respond rapidly to customer requests for a single data capture solution for a variety of document types and formats.

The KTP has delivered an innovative deep learning model for natural language processing (NLP) that analyses the content and structure of a range of digital documents – including invoices and tax forms – sorting the information into categories for ease of use.

This improved system will streamline a range of financial processes for Xceptor customers worldwide including opening bank accounts, approving mortgages, responding to customer queries, and managing insurance claims.

This means that Xceptor’s clients can now administer routine services more efficiently, freeing staff to concentrate on higher value tasks.

In addition to the new product line, the KTP has also enhanced the organisation’s project management, people management and technical leadership skills – with Chris Smith, the project’s company supervisor, now leading a growing AI team at Xceptor.

In his new role, Chris will continue to work with the project’s KTP Associate, Dr Chao Zhang, who now works with Xceptor as an AI Engineer, embedding his expertise within the company.

Indeed, having provided Xceptor with a deeper understanding of the demands of AI systems, the project has prompted a re-organisation of its development teams to support ongoing refinement of machine-learning models and enhance the user experience.

What’s more, the KTP has created new data analytics capabilities and forged new communication channels between Xceptor presales and software development teams – supporting further R&D pipelines and innovation.

“In addition to the technical achievements of the KTP – which have created significant new commercial opportunities for Xceptor – working with the academic team has allowed us to build the skills we need to develop robust and innovative AI capabilities independently. We've also found a fantastic partnership with the University which we hope to continue long after the KTP."

Chris Smith - AI Technical Lead at Xceptor

Benefits for the University

The existing relationship between the University and company has been strengthened in multiple ways – not least with Chris Smith’s new membership of the University’s Industry Advisory Board.

Over the coming years, Xceptor is keen to support further collaborative research in this area via PhD studentships and placement projects as well as R&D funding applications to Innovate UK. In addition, they hope to grow their new Data Science team – recruiting from the University’s AI and Computer Science graduates.

Both undergraduate and postgraduate student learning has been enriched with teaching materials arising from the partnership – including guest lectures, a new NLP module for Masters students, and related projects in NLP. All of this new activity provides students with important vocational insights for their future careers.

The academic team have acquired invaluable experience through applying their knowledge to real-world challenges. They have also expanded their research portfolio with an IEEE conference publication and journal paper for submission in a high impact research journal.

Discussions are already underway between the Xceptor technical team and academics at Loughborough as well as Keele University to uncover further areas for collaborative research. Several topics with wide application impacts have been identified – including rapid automated data extraction to accelerate checks of fraudulent activity.

“This has been a highly successful KTP. The team have delivered commercially relevant outputs that enhance Xceptor’s offering whilst advancing the academic research agenda.”

Dr Matt Hogan – Knowledge Transfer Adviser at KTN

Meet the experts

Photograph of Baihua Li

Professor Baihua Li

Professor of Computer Vision and Machine Learning

Photograph of Eran Edirisinghe

Professor Eran Edirisinghe

Pro Vice-Chancellor (Research and Innovation) Keele University

The UKRI Innovate UK logo

Knowledge Transfer Partnerships aim to help businesses improve their competitiveness and productivity through the better use of knowledge, technology and skills within the UK knowledge base.

This KTP project was funded by UKRI through Innovate UK.