The contrast between the old personal-touch era of banking and today’s fast-paced digital world is striking.

Customers once relied on familiar local branches for tailored advice and personal service. Now, as increasingly global and time-pressed consumers, they expect instant, seamless interactions.

Crucially, their benchmarks are no longer set by rival banks but by digital leaders like Amazon, which have redefined what a frictionless experience looks like – for better or worse (there’s an argument to be made that having a bit of ponderous friction for ‘buy now, pay later’ has its own benefits).

Regardless, to keep pace, financial institutions must move beyond being transactional utilities and instead act as trusted partners who guide customers through major life events – a shift that demands both long-term vision and steady, incremental improvements.

In Edinburgh at DIGIT’s annual Fintech Summit, ThoughtWorks’ Sapna Maheswari ran with these themes in her keynote talk, setting the tone by challenging banks to rethink how they apply AI.

Speaking to a packed auditorium at Edinburgh’s EICC, Maheswari approached the issues of meaningful AI integration from a decidedly CX-coded perspective, cautioning against quick-fix tech upgrades and emphasised the need for banks to address core process issues first.

Her keynote highlighted how soaring consumer expectations – shaped by seamless digital services – now force financial institutions to rethink their business models to stay competitive.

Let’s get into it.

Fix Your Processes Before Automating

Maheswari’s first message was a simple one: start with the basics.

AI should be used to enhance core services – like personalised recommendations and fraud prevention – but only on well-designed processes. In one example she described a bank overwhelmed by thousands of customer service requests.

According to the case study Maheswari presented, analysis showed “almost half of those requests were coming in because the teams couldn’t retrieve the information they needed,” leading to unnecessary escalations.

The bank had tried to automate its service processes, but “people are looking to fix processes or automate processes that are not necessarily the right thing to fix,” Maheswari observed. The lesson, she said, was to step back and map the customer journey before deploying any solution.

She gave another cautionary example around chatbots. Maheswari warned that without solid data and decision logic, even a sophisticated AI assistant can make things worse.

She said: “If you don’t have the data and decision-making process in place, the chatbot just becomes a solution that potentially deteriorates the customer experience,”

Giving an example she experienced herself, Maheswari described asking a bank’s bot to recommend the best credit card for her, only for it to reply “Would you like to know what your balance is for?” – a useless answer that “takes the confidence down” of the customer.

Maheswari emphasised that banks must always think about experience – and the underlying processes – first, not just the shiny new technology.

Rethinking Products and Culture

Once the basics are in order, Maheswari said, companies can begin reimagining their services.

She shared a life-sciences case study: a pharmaceutical client needed to speed up drug discovery. ThoughtWorks’ team first “built a data platform to put together disparate clinical data from various sources, to create the foundation.”

On that foundation they added AI tools: they built “a research co-pilot assistant” to help scientists query the data conversationally, and then “a multi-agent system to allow the researchers to create regulatory reports,” she explained. By focusing on one narrow sub-process, they were able to prove AI’s value and then expand outward.

Two key lessons emerged.

First, a layered approach – strong data platforms underpinning AI tools – lets companies progress steadily while minimising risk.

Second, success breeds excitement: as more employees see these AI tools in action, the organisation naturally evolves toward becoming “an AI-first people culture.”

Maheswari urged firms to “sow the seeds” by giving staff hands-on AI experiences, since the more people see these tools working, the more they will embrace them for their customers.

Agentic AI: The “do-it-for-me” Economy

Looking further ahead, Maheswari described a future of agentic AI – systems that take multi-step actions on behalf of users without their oversight.

In this scenario, customers might stop asking “What are the best mortgage options?” and instead have an AI agent that manages renewals automatically.

As she put it, “we’re not talking about chatbots… we’re talking about agents that will take decisions on your behalf.” For example, a banking assistant might ultimately decide “I’ve got my mortgage renewing in December, can you find the best options for me… and do the switch.”

Such a proactive scenario may sound kind of preposterous today given the trust element with AI just isn’t there on the consumer side (something Maheswari explores shortly) to allow to make such big decisions, but Maheswari said ThoughtWorks is already exploring it.

In particular they are experimenting with future interfaces that blend traditional menus and icons with conversational, generative elements that adapt in real time to context and user needs.

Humans, Trust and Closing the Lab-to-Life Gap

Despite the focus on AI, Maheswari stressed that humans remain the industry’s most important asset.

“We’re not dealing in data and transactions, we’re dealing in trust,” she reminded the audience.

For example, a customer who’s been defrauded or lost savings is “looking for reassurance” from a human advisor.

AI can support people, but it can’t replace the empathy and judgment customers need in those moments. Likewise, humans are essential for overseeing AI itself: governance, compliance, security and ethics all require human judgement.

As Maheswari put it, “This is not about replacing people – it’s about augmenting them so you create better experiences.”

She also warned against letting projects languish in the so-called “lab-to-life gap,” where promising pilots never reach customers.

To bridge that gap, firms must start with a clear customer problem (never deploying AI “for the sake of the solution”) and ensure their data and technology foundations are solid.

Recommended reading

• Fintech Summit 2025 | Less Than a Week To Go!
• UK Gov Looks to Cut Red Tape for Fintech Innovation
• Are Fintechs Entering a New Golden Era?

“AI is nothing without the data or technology foundations,” as Maheswari puts it.

Companies should then make bold strategic investments in AI and stick with them – in her words.

On this, Maheswari said: “Make bold bets in the business and stand behind it.”

This means breaking down silos so that all teams – including compliance, security and risk – work together on innovation.

“We need to think about security, policy, ethical considerations very, very differently,” she added.

In closing, Maheswari emphasised collaboration and experimentation. No one has all the answers, she said, but the industry will learn by trying new ideas together.

“Let’s shape it together,” she urged.