By Rajiv Desai
AI using Natural Language Processing (NLP) and Machine Learning (ML) can significantly help improve payment processing by automatically performing payment auto repair, enrichment and intelligent routing – just as an experienced human payment operator would..
There have been many recent conversations about the use of Artificial Intelligence (AI) technology in the banking industry, especially in relation to retail banking and brokerage. Less explored are the ways that AI can significantly impact wholesale banking and especially on high-value and cross border payment processing.
There remain many inefficiencies in processing these types payments, due to the levels of manual intervention that are costly and time consuming. These inefficiencies are due to incorrect, incomplete or misplaced information regarding the payment, often a result of lack of knowledge or the legacy applications that generate these payment instructions. Incomplete payment instructions cause delay in processing, increase the costs, and result in a failure of SLA and negatively impacting customer satisfaction.
AI using Natural Language Processing (NLP) and Machine Learning (ML) can significantly help improve payment processing by automatically performing payment auto repair, enrichment and intelligent routing – just as an experienced human payment operator would. There is still lot of free format descriptive information provided in payment instructions, and we expect this will continue when ISO 20022 or MX message formats are used.
As corporates are not going change their ERP and other applications, they will continue to send existing payment instructions, with Bank applications being forced to map legacy message format into the new ISO 20022 or MX format without much value add.
Using Natural Language Processing techniques, we are able to correctly parse, identify and normalize free text information into series of well-defined data elements that computers can understand. This enables systems to derive proper bank codes (such as BIC, FED, CHIPS, Sort code, BSB codes etc.) then format these correctly into various appropriate payment fields either in SWIFT or ISO 20022 formats.
Once data is normalized then it can be further enhanced to perform additional enrichment that is targeted for a particular bank or country or currency or back office specific instruction to further improve STP rate.
Auto-Repair and Routing
AI & NLP also allows reviewing information across several fields to perform intelligent intra-field auto repair such as main bank information in one field, while branch information that is in a completely different field. The same techniques can be further leveraged to enable least cost routing based on the entire content of the payment, to provide timely payment processing without any human intervention.
This AI-powered approach allows banks to reduce manual intervention cost, improve operational efficiencies and better customer service. Banks around the world using AI tools have achieved higher than 97% STP rate and significantly improved margins. For any organisation looking to improve STP rates, AI technologies should be at the heart of your strategy!