EXCLUSIVE: “Beyond The Cloud” – Karen Zhang, Google Cloud in ‘The Paytech Magazine’

Google Cloud’s Karen Zhang maps out the many ways in which the infrastructure provider is helping to build a modern financial services community

It’s the world’s third largest Cloud technology provider, but Google Cloud has emerged as much more than a convenient host. The company works closely with fintechs and financial institutions, not only supporting them from a Cloud infrastructure perspective – so they don’t have to worry about ‘keeping the lights on’ and can instead focus on developing the products and services for customers; it’s also helping guide them through an age of rapid regulatory change and accelerated innovation.

We spoke with Google Cloud’s Start-ups Lead for UK & Ireland, Karen Zhang, to understand how and why the technology giant is investing time and money in the sector.

Her job, she says, is about more than technology development and deployment: “We also develop cultural ties [though hackathons and workshops]. It’s exciting to work with early-stage fintechs from the ground up and see them grow.” Here are four distinct ways in which she believes Google Cloud is a game changer for its partners.

Creating an optimal operating environment

New regulations create new considerations. The European Union’s Digital Operational Resilience Act (DORA), for example, requires fintechs to have high-level redundancy and resilience measures in place for critical functions to ensure business continuity. Google Cloud is working with its partners to help them demonstrate multi-Cloud strategies and robust security practices to meet the requirements of the Act.

“We support companies’ core platform services to ensure they’re compliant and resilient from a security standpoint,” says Zhang. “That’s super-critical and one way in which we can really help them at the foundational level.”

She draws an analogy with an environment in which risk is also a critical issue. “If you think about Formula One racing, you can’t have the confidence to really accelerate if you don’t have your brakes in place.

“We see security as the brakes when it comes to Cloud infrastructure.”

Moving beyond the structural foundations, Google Cloud also helps fintechs develop their product propositions, particularly around personalisation and explainability, which is subject to the EU AI Act’s responsible use rules. Zhang likens Google Cloud’s approach to joining Lego bricks. It starts with the infrastructure, then builds in data and AI tools, which unlock deeper insights that drive value-added products and services.

“One good example is the work we’re doing with the UK’s Starling,” says Zhang. “We’ve helped it create a spend intelligence service, so users can use written or voice prompts to check their spending behaviours. They can ask, for example, ‘how much have I spent on transport over the last week and how has that changed, week-on-week?’ It enables them to better engage with, and understand, their financial behaviour, and that’s powerful.”

Google Cloud also supports back-end services. It helped embedded finance specialist Liberis build its AI underwriting agent Ada, for example. Underwriting involves large volumes of data and repeatable steps, creating a heavy load for smaller fintechs in particular. Ada does the lifting for them.

“By working in conjunction with their underwriting team, Ada’s reduced overheads by 50 per cent, allowing them to focus on more critical, knowledge-based work,” says Zhang.

Conquering compliance with explainable AI

Google Cloud is working with leading financial institutions to redefine fraud prevention using AI. Its Anti Money Laundering AI, based on advanced large language models (LLMs), launched in 2023 an has already helped banks and fintechs to identify more risk, more defensibly, with fewer false positives and reduced time per review.

This includes partnering with HSBC to co-develop Dynamic Risk Assessment, the AI system that the bank uses to check for financial crime.

“In the past, they relied on a legacy AML solution that was quite rudimentary and restricted to a rigid rules-based system to detect anomalies. It was coming back with 95-98 per cent false positives,” says Zhang. “So we worked with them to leverage [previous iterations of] Gemini and some of our explainable AML solutions to increase their fraud detection rate by two to four times, while reducing false positives by 60 per cent.”

The bank no longer needs to call so many customers unnecessarily to ask them about what turns out to be completely legitimate activity.

“This has resulted in a massive reduction in terms of the overheads and human costs involved,” adds Zhang. “There is still very much a human in the loop; it’s just made their job more efficient.”

The UK’s Starling, meanwhile, has adopted a new purchase scam intelligence tool, modelled on the latest Gemini multimodal large language models (MLLMs). Applying MLLMs to compliance processes can help firms better understand both text and visual data, such as images and documents, to automate, monitor, and enforce regulatory requirements. Unlike standard LLMs, which are limited to text, MLLMs can also verify visual evidence.

In the case of Starling, it provides customers with a way to upload images of items, marketplace ads, and even messages from sellers for the tool to analyse in a matter of seconds for potentially fraudulent activity. It is the first of its type in the UK.

Making the most of MLLMs

Released in November 2025, Google Cloud’s Gemini 3 is a multimodal large language model (MLLM) that has opened up new use cases that derive value from unstructured data. These advanced AI systems that integrate text-based LLMs with sensory encoders to perceive reason, and generate content across multiple formats, including images and video, draw on wider sources to fuse perception with reasoning.

To illustrate Gemini 3’s potential, Zhang provides an example from a customer that manages around 20 per cent of the UK’s fleetmanagement services, working closely with Amazon Logistics. Traditionally, producing its driver safety training videos was a costly affair.

“You’d have entire film crews and would need to block off a street for an entire day,” she explains. “Imagine the overheads needed to generate this content.”

But using Google Cloud’s video generation model (Veo 3) alongside Gemini 3, the team now generates safety training content through prompting, which has resulted in more than 50 production-grade videos.

What was once labour-intensive and expensive has become faster, more flexible, and more scalable.

“A year ago, this would have been hard to imagine because we were still in a space that was so focussed on structured data,” adds Zhang. “Whereas now we have all this unstructured data from videos and images that we can actually use to create meaningful, impactful content.”

Thinking from a financial services perspective, Gemini 3 can instantly comprehend complex text, images, video, and audio files to provide fintechs with far more usable information from unstructured data. And this informs better decision-making and more personalised customer experiences.

Scaling ambition

Whether it’s streamlining back-office processes or curating customer personalisation, Google Cloud is helping fintechs and financial institutions scale efficiently. In Zhang’s experience, AI liberates founders who previously may have been constrained by the challenge of securing investment in an impossibly long development runway.

“It’s opened doors and enabled many different founders, bringing them to a place where they don’t need all this venture capital backing, which has typically stopped many people with great ideas from getting a head start,” says Zhang.

It’s also underpinning the transition from B2C to B2B revenue streams, which is an increasingly prominent strategic shift, driven by the need for profitability, lower customer acquisition costs, and more predictable revenue models. With B2C fintechs facing high churn, tight margins, and intense competition, many are pivoting to offer ‘enabling’ services, leveraging their existing technology to serve other businesses.

“They’re essentially white-labelling services and building out certain components of their tech stack for larger enterprises,” explains Zhang. “We’re seeing a huge shift from B2C to B2B ‘banking-as-a-service’ propositions, and that’s where I’m spending a lot of my time – supporting those fintechs that have shifted their business models.

“It moves the conversation from technology to supporting new go-to-market strategies and a whole new level of scale where founders can deploy their proposition across millions of users by expanding into the Tier One ecosystem.”

This article was published in The Paytech Magazine Issue #18, Page 14-15

The post EXCLUSIVE: “Beyond The Cloud” – Karen Zhang, Google Cloud in ‘The Paytech Magazine’ appeared first on FF News | Fintech Finance.