NEW YORK — Fintech Finastra is working with clients to deploy agentic AI within payments processes to boost efficiency and accuracy. 

The London-based company is collaborating with clients to identify what repetitive processes can be automated or streamlined through agentic AI, Chief Executive Chris Walters told Bank Automation News at the firm’s Americas Lending Day event Sept. 18.

In the payments business, for example, trillions of dollars are sent globally every year, which involves payments operators, said Walters, who became CEO in January. These operators — thousands of them at some banks — could mis-key information in the transaction process.  

Payments operators are responsible for managing and processing financial transactions, which include inputting correct payee and payer information, according to payments provider Modern Treasury.  

To avoid mistakes and to operate at scale, payment operators can lean on AI for: 

• Correction suggestions; and 

• AI validation up front. 

“That could reduce a heck of a lot of labor,” he said. 

Lessons learned 

The human element of any operation can result in errors — sometimes significant. 

In February, for example, a payment operator at Citi mistakenly initiated a payment of $81 trillion instead of $280. 

This mistake, which was caught and reversed before the money left the bank, was an eye-opener for “better technical controls around their payment authorization policies,” James McCarthy, founder and chairman of financial consulting agency McCarthy Hatch and a founding member of the Consumer Financial Protection Bureau, told BAN at the time. 

Ronak Doshi, partner at consulting firm Everest Group, told BAN in February that agentic AI tools can help avoid such events by performing “multiple validations and review client work and flag unusual activity that is not previously observed.” 

AI approach 

Along with agentic AI exploration, Finastra is also deploying AI within internal and external operations, Walters said, noting a “top-down and bottom-up approach.” 

In the top-down approach, management looks at repetitive organizational, Walters said. With the bottom-up approach, Finastra is giving all employees AI tools to determine where they can have success with the technology, he added. 

Current AI uses at Finastra include: 

• Coding and testing; 

• Chatbots for customer support; and 

• Lending assistance. 

Finastra has deployed Microsoft Copilot to provide employees with gen AI tools and is using LLM model Cursor for coding purposes, Walters said, adding that it is taking a multimodal AI approach because it doesn’t want to be locked into one offering. 

The $434 million First Pacific Bank, $665 million FLC Bank, $1.5 billion Beacon Credit Union and $1.1 trillion Lloyds Bank are Finastra clients.