Four weeks to build Scam Intelligence tool
50% reduction in customer service referrals
Created intuitive Spending Intelligence tool with Gemini
Offers AI capabilities to other banks with Google Cloud
8,000 hours saved per month
Starling uses Google Cloud AI to help customers with their money, while empowering banks to build AI solutions of their own.
Starling uses Google Cloud AI to help customers with their money, while empowering banks to build AI solutions of their own.
Bold AI ambitions built on a solid data foundation
We lacked the infrastructure to support the necessary data science work. As we looked for an alternative, it was clear Google Cloud was the right provider due to its experience in ML. We decided to move our data warehouse to BigQuery and build our ML training capabilities with Google Cloud.
Frédéric Laurent
Deputy CIO, Starling
Starling's mission is simple: to help customers be good with money. Part of Starling Group, and one of the UK's first digital banks, Starling puts technology at the heart of everything it does, giving customers new ways to understand their finances so they can spend, save, and manage their money effectively.
A digital native, Starling built its core platform on cloud infrastructure. However, as the Starling team made increasing use of machine learning (ML) to develop customer solutions, they realized their existing data and ML stack were a bottleneck to development. With its previous stack struggling to handle the load of a growing number of data producers and data consumers, Starling needed data capabilities that could handle the bank's significant ambitions in AI and ML.
"We lacked the infrastructure to support the necessary data science work," explains Frédéric Laurent, deputy CIO at Starling. "As we looked for an alternative, it was clear Google Cloud was the right provider due to its experience in ML. We decided to move our data warehouse to BigQuery and build our ML training capabilities with Google Cloud."
Along with migrating its core data processing to BigQuery, Starling adopted Vertex AI for model development. Vertex AI Workbench was key to its decision, offering a secure, controlled environment that let the bank’s data scientists iterate fast without risking data exfiltration, meeting the critical security requirements of a regulated bank.
Helping customers take control of their money with Gemini and Vertex AI
With BigQuery and Vertex AI, Starling could combine its wealth of data with powerful ML capabilities to rapidly deploy sophisticated AI features. Now, with Gemini models, Starling is accelerating that innovation. The bank recently built a feature using Gemini's powerful multimodal capabilities to actively protect customers from scams.
The tool, Scam Intelligence, allows customers to submit a picture of a marketplace ad, such as one from Facebook Marketplace. Gemini processes this information, looking for tell-tale red flags, such as a suspiciously low price or pressure tactics from the seller. This enables the customer to make an informed decision as to whether or not the ad is legitimate or a scam, and teaches them the warning signs to look out for in the future. Scam Intelligence has already increased the rate at which customers cancel marketplace payments by 300%.
We built our Scam Intelligence feature really quickly thanks to Vertex AI. With the ML pipelines and Vertex AI Workbench, we codified the entire model training process from the start, allowing the training code to run without manual intervention, enabling us to deliver the first phase of the tool for consumer testing in just four weeks.
Frédéric Laurent
Deputy CIO, Starling
The specific Gemini prompt used to analyze this information is continuously evaluated and managed using Vertex AI Pipelines, ensuring the bank's intelligence is always accurate and up to date. Meanwhile, the Starling team has used Model Armor to build guardrails around the solution so it can only be used as intended.
"We built our Scam Intelligence feature really quickly thanks to Vertex AI," recalls Laurent. "With the ML pipelines and Vertex AI Workbench, we codified the entire model training process from the start, allowing the training code to run without manual intervention, enabling us to deliver the first phase of the tool for consumer testing in just four weeks."
Starling is also using Gemini to help customers make more informed spending decisions with its Spending Intelligence tool. They can simply ask questions about their money using natural language, such as, 'how much did I spend on coffee last week?' Gemini then quickly converts the question into a structured function call that triggers Starling's analytics engine, helping to ensure the customer always gets an accurate response. This makes it easier for them to understand their finances in an intuitive way, without having to interact with filters and graphs.
Meanwhile, Starling is using Gemini to save its customer service team time that was previously spent manually summarizing customer call transcripts. By automating the task with Gemini, the bank helps save 8,000 hours per month. Starling has also implemented Gemini into its help center to confirm customers have the information they need when using the app. By feeding Gemini with all the bank's customer-facing content, Starling offers instant, accurate answers to complex customer queries, helping customers move beyond frustrating keyword searches, and reducing the number of customer service referrals by 50%.
Scaling the AI blueprint for digital banking for all
Underpinning any AI solution is the need for a robust, modern data foundation, which is how we've architected the Engine platform from day one. By leveraging our technology and everything we’ve learned with Google Cloud, we can help clients build and deploy generative and agentic AI solutions safely, at scale.
Krishna Aswani
Regional Manager - North America, Engine by Starling
Now, Starling Group is offering its innovative capabilities to other banks too. With Engine by Starling, Starling is providing its modern, cloud-native core banking platform to banks around the world, helping them achieve the economics and efficiencies that Starling has realized.
By providing banks with a proven platform and the operational blueprints to succeed, Engine by Starling helps banks de-risk modernization and meet the rapidly evolving needs of their customers. Thanks to the platform's highly-intuitive data model and event-based architecture, Engine by Starling’s client banks can offer AI-powered customer services, enhance their risk and compliance capabilities, and leverage AI as an additional tool to solve business problems and meet customer needs.
"Underpinning any AI solution is the need for a robust, modern data foundation, which is how we've architected the Engine platform from day one," explains Krishna Aswani, Regional Manager - North America at Engine by Starling. "By leveraging our technology and everything we've learned with Google Cloud, we can help clients build and deploy generative and agentic AI solutions safely, at scale."
Engine by Starling helps client banks across three key areas: Firstly, it enables banks to pursue both the growth and efficiency that Starling itself has realized as a highly profitable, digital bank.
Secondly, the platform unifies both the customer and employee experiences. It provides a real-time, 360-degree view for customers to self-serve across all their banking relationships, while empowering bank employees - and even AI agents - to deliver highly personalized customer experiences.
Finally, the platform provides banks with the modern foundations, context, and tooling to enable the agentic workforce of the future. With Engine's API-first and MCP-ready capabilities, banks can experiment with and deploy AI agents alongside their human workforce to run a compliant and highly effective operation.
For Laurent, that vision is fast becoming a reality thanks to Starling's work with Google Cloud data and AI services. "Google Cloud has brought us enterprise data and AI capabilities where previously we were limited in what we could achieve," he explains. "As a result, we've been able to quickly build powerful, innovative features for customers and staff alike."
One of the UK's first digital banks, Starling aims to help customers manage their finances effectively. With Engine by Starling, it's helping other banks modernize to meet the needs of their own customers.
Industry: Financial Services
Location: UK
Products: Google Cloud, BigQuery, Gemini, Model Armor, Vertex AI
