A brownie is an undemanding customer. It asks for no mortgage, needs no overdraft and rarely calls the contact centre. In October 2016, however, a £1.80 brownie provided a rather exacting examination of the machinery Sam Everington and his colleagues were building at Starling.
Anne Boden, the bank’s founder, had taken its new debit card to London’s Regent Street. Back at headquarters, the team watched for the transaction. Months of work had to survive an ordinary purchase. The payment appeared on their screens. The card worked. The relief, Everington later recalled, stayed with the people watching.
There are grander ways to describe a career in banking technology. This small purchase has the advantage of explaining the job. A customer buys something; an elaborate collection of systems must agree about what happened. The customer should be free to think about the brownie.
The engineer on both sides of the counter
Everington joined Starling in 2016 to build the systems and apps behind its current accounts. His work expanded into leading the design and development of the banking platform itself. He now runs Engine by Starling, the business that supplies that technology to other financial institutions. London remains his base, while his customers have taken the software much farther.
His earlier work was less visible than a banking app. He began as a software engineer developing an accounting package, with responsibilities stretching through sales, requirements, process design, implementation and support. That is a useful education in consequences. A feature can look finished to its developer while leaving the person who must use it with six new questions.
At Starling, engineers worked alongside the people operating the bank. Customer service, payments and financial-crime teams could feed problems directly into development. Everington has described the importance of that proximity: staff could try something, discover its shortcomings and have it adjusted. The loop between a complaint and a change became shorter.
That experience helps explain his interest in the employee’s screen. A banking app gets photographed; the tools used to answer a customer’s query generally do not. Yet a smooth customer experience depends on someone being able to understand the account, find the relevant information and act. A handsome app cannot do the whole shift.
Eight weeks before yes
When he was asked to lead the separate software business, Everington did not immediately seize the opportunity. He took about eight weeks to agree. He loved his existing job and needed time to decide whether he was ready to leave it. Ambition, in this instance, had to negotiate with enjoyment.
The eventual move put an engineer in charge of a commercial proposition with an awkward requirement: convincing other banks to trust machinery developed inside somebody else’s bank. Admiration is pleasant. A commitment to run your business on the thing you admire is considerably more demanding.
Engine began in 2022 with Everington and chief product officer Victoria Newton as its two-person team. The opportunity had been arriving at Starling’s door. Visiting executives wanted to understand how its technology and operating model worked, and conversations kept turning into requests to collaborate. Office visits had become a form of market research.
By April 2019, Everington was already speaking publicly as Starling’s head of engineering about open banking. His concerns included useful customer propositions, straightforward authentication and secure data. Those are revealing priorities for someone whose later job would involve selling technical change. The test begins with whether a person can use the thing.
“I’m an engineer at heart”
Sam Everington, February 2026
The rule that travels with the software
Engine’s offer extends through the bank: accounts, payments, customer journeys and the tools staff use to run operations. The platform is supplied as a managed service. Hosting, maintenance and ongoing upgrades sit within that arrangement. A client is buying access to an operating system and its continued development, rather than simply taking delivery of code.
Everington has been firm about the economics of keeping that product shared. New capabilities should be available across the banks using it. Local requirements still need configuration and integration, but allowing every client to acquire a private software branch would change the business. The attractions of a common platform would become harder to preserve.
For a bank, this asks a serious organisational question. How much of its existing procedure expresses something valuable about the institution? How much survives because an old system once required it? New software makes that distinction uncomfortable. A familiar workaround can develop the social standing of a family heirloom.
Everington’s commercial challenge therefore reaches beyond engineering. The people buying Engine must see a reason to change how work gets done. In a conversation with Mark Bernhardi in Sydney in July 2026, he described partnering with established banks that bring their own brand, licence, customer trust and local knowledge. Each side contributes something the other would take time to build.
Romania puts a date on the promise
Salt Bank gave the proposition an early public examination. Its team wanted to launch a digital bank in Romania, and the timetable was about a year. For Everington, who had helped Starling reach its own launch, this offered a chance to repeat the exercise with a system already built.
The work was still local and specific. Engine configured the platform for Romanian requirements and embedded staff in Bucharest. GFT helped deliver the app and integrations; AWS supplied the cloud foundation. Salt launched in April 2024, with accounts in Romanian lei and 16 foreign currencies, card controls and savings spaces.
Its choice of service model also matters to Everington’s story. Salt kept real people available through a round-the-clock care centre. Employees used a shared view of the customer in Engine’s management portal. The digital bank still needed a human who could help, equipped with enough information to be useful.
After a year, Salt had 500,000 customers. Customer acquisition belongs to the bank and its team; it is not a personal trophy for the software supplier’s chief executive. For Everington, the result nevertheless supplied something a demonstration cannot: evidence of people opening accounts and using a bank outside the UK.
Public launch, Romania
Public launch, Australia
Launch, Canada
A second first payment
Australia supplied a familiar scene with a larger audience. AMP Bank chief executive Sean O’Malley used a new AMP debit card to buy a team lunch at a charity event. Colleagues cheered, and the transaction was filmed. Everington recognised the anticipation from London in 2016. This time there was considerably less privacy if lunch failed to materialise.
The project had started in October 2023. Its first production accounts went live in October 2024, before the public launch in February 2025. That distinction is useful: building a working bank within a year and making it available to the general public are separate milestones.
AMP Bank GO’s first year then brought the less theatrical work of extending the product. Lending, term deposits, joint accounts and integrations with Xero were among the additions. For Everington, the ongoing relationship is central to the offer. A launch proves that a system can run. Subsequent changes test whether the bank can keep improving it.
Keeping a person in the conversation
Artificial intelligence gives another view of his priorities. In 2023, Everington favoured using AI in the background to improve customer journeys. He was unconvinced that putting it in front of a customer automatically produced a more enjoyable experience than a simple app journey or a conversation with a person.
By a July 2026 interview in Berlin, he was discussing a wider role: humans checking AI actions, and AI spotting things humans miss. His argument centred on comparing outcomes and catching errors. Both parties could overlook something. Efficiency mattered partly because it could leave people more capacity to serve customers.
He also described practical examples, including help assessing suspicious payment-related images and an assistant able to guide customers through actions such as creating savings spaces or freezing a card. His examples stayed close to things customers already need to do. The technology had to earn its place in the transaction.
The next test has a Canadian address
On 5 October 2026, Tangerine launched its new business banking proposition on Engine, less than a year after signing the agreement. The Canadian bank is owned by Scotiabank, and the launch marked another international application of software that began inside Starling. The initial offering was a business proposition; it should not be confused with the completed migration of every Tangerine customer.
New Zealand’s SBS Bank represents a different stage. It announced its Engine partnership in February 2026, with Deloitte supporting the transformation. SBS told members to expect customer-facing changes from late 2027. A signed contract starts work. Customers experience the result later.
Everington’s stated ambition now includes £100 million in annual recurring revenue for Engine. That is a target, rather than a reported result. He also wants to demonstrate that banks using the platform can combine customer satisfaction with efficient operations. The claim will be tested across different institutions, with different customers and different demands.
The distance from a £1.80 purchase to that ambition is substantial. His own accounts of the work keep returning to smaller events: a payment appearing, an account going live, a customer getting something done. For an engineer running a banking software business, these are useful things to remain excited about. Someone taps a card. Lunch arrives. The machinery has earned another day.