HSBC’s SAS Viya pilot puts intelligent banking’s foundations to the test

The future of AI in banking hinges on whether banks can use AI to improve decisions while retaining oversight and accountability.

HSBC’s extended relationship with SAS, announced on 5 October, offers a useful starting point. The relationship dates back to 1998. HSBC now plans to combine its internal technological capabilities with SAS expertise to pilot the Viya platform.

AI in banking needs more than faster answers

When most people think about AI in banking, they imagine interacting with a customer service chatbot. But the opportunity for banks extends beyond chatbots to heavy-duty tasks like catching fraud, stopping money laundering, boosting cybersecurity, and sifting through massive amounts of data to catch trends. That is what the Bank of England and FCA’s 2024 survey, covering 118 financial-services firms, found.

However, 46% of respondents using or planning to use AI reported only a partial understanding of their AI technologies. Third-party models were a particular source of that uncertainty.

For banks, the challenge is turning analytical capability into decisions that can withstand scrutiny. After all, bankers can’t just tell a customer “the computer told us to,” after denying them a loan. They have to be able to prove why the decision was made, both to their customers and to government regulators.

This aligns with our coverage of Acuity Analytics’ financial-services AI platform where we argued that credit analysis, regulatory summaries and investment research need traceable outputs and informed human review, not merely faster production.

From platform pilot to operational proof

A useful pilot should establish what improves, what can go wrong and who remains responsible.

Consider a hypothetical risk analyst receiving an AI-generated assessment. Its value depends on whether the analyst can inspect the underlying evidence, identify missing information, and challenge the conclusion.

Then comes integration. 

A recommendation must reach the right workflow, with appropriate access restrictions and escalation routes.

Our examination of how ARIS connects sovereign AI with business processes highlighted that distinction: where AI runs and how it exercises authority are separate questions.

For HSBC and SAS, the next meaningful evidence would be a defined use case, acceptance criteria and results measured against an existing process.

The prize is confidence: better-supported decisions, defensible risk assessments and less operational friction.