In its first vertical industry-specific generative AI release, Salesforce took a big swing today by going after one of the toughest customer audiences it serves: Financial Services Cloud users.
Not only that, but Salesforce also applied AI to one of banking’s most difficult problems to automate: transaction disputes.
GenAI-powered Transaction Dispute Management comes with prompts that generate text for various phases of the process, such as dispute acknowledgment, opening, resolution and closure. It also includes Service Processes Library, which now includes about 20 prebuilt processes — dispute management being one — and more to come, said Greg Jacobi, Salesforce vice president and general manager of banking and lending.
Salesforce also plans to release a chatbot called Einstein Copilot Banking Actions this fall, as well as a Copilot Studio process design tool with which users can build their own service processes.
It’s a gamble to roll out GenAI to banks, said Rebecca Wettemann, founder of Valoir Research. Salesforce Financial Services Cloud users are typically in highly regulated industries — with tangled processes designed to comply with state and federal regulations, PCI standards, consumer protection laws and their own high privacy, security and service standards. From a technological perspective, banking customers’ data may be siloed both on-premises and in the cloud.
As such, banks aren’t typically early adopters of bleeding-edge technology. In theory, they should be less likely to trust GenAI; their customers are targeted by hackers and fraudsters who use false identities to access their systems. When a transaction dispute resolution goes wrong, two bad things can happen:
The customer leaves for a competitor; or
Bad actors steal from the enterprise.
Using AI as part of a transaction dispute process can be risky but worth the reward if it works well for financial institutions.
If you can get [GenAI] done in financial services, it’s going to work everywhere else because you’ve cracked the toughest nut. Rebecca WettemannFounder, Valoir Research
“A year ago, at the Salesforce Connections conference, everybody was talking about how you’ll never be able to do AI in financial services because you won’t have the audit trail and compliance stuff when you have AI doing it,” Wettemann said.
“I thought it was very interesting that financial services is the first area where Salesforce introduces vertical-specific [GenAI] capabilities,” she said. “If you can get it done in financial services, it’s going to work everywhere else because you’ve cracked the toughest nut.”
Banking compliance can be thorny
Despite the skepticism among financial services workers toward AI in general, Salesforce has found interest in GenAI among those customers for several reasons, Jacobi said. First, GenAI can access more data from policies, procedures, history and knowledge bases and synthesize an answer more rapidly than human agents. So as a co-pilot for human agents, the technology can suggest more precise solutions to customer problems.
Second, with Salesforce’s process transparency and data protections within the Einstein Trust Layer, banks can trace how the AI arrived at a suggestion, and also keep customer data where it needs to stay for compliance.
Compliance officers — often the toughest audience to please in a financial services enterprise — appreciated how GenAI can absorb and invoke all of a bank’s policies and procedures when tackling customer problems during the beta testing process. In fact, Salesforce’s GenAI found previously unknown conflicts among policies and procedures at one bank as it trained its AI model, according to Jacobi.
The Salesforce GenAI bots for financial services are currently in “co-pilot mode,” as agents will need a human in the loop for some time to come. But Wettemann pointed out that if a bot can handle transaction disputes in an automated manner, financial institutions could potentially bank more profits.
“Fraud dispute resolution is a huge expense for banks, and it’s a huge amount of frustration for customers,” she said. “It’s an area where you’ve got to have a certain level of intelligence, but you also want to have consistency.
“It’s hard to train, onboard and keep agents in those roles. So, if I can do it completely self-service, great. If I can’t — but I can give agents tools and capabilities, so they stay longer because they’re less frustrated — it’s a win.”
Salesforce released its Financial Services Cloud GenAI tools in conjunction with Salesforce World Tour NYC. Pricing information can be found on Salesforce’s site.
Don Fluckinger is a senior news writer for TechTarget Editorial. He covers customer experience, digital experience management and end-user computing. Got a tip? Email him.
