They Rebuilt a Bank Like a Startup - Here's What Actually Worked

Here's a weird origin story: in 2020, Richard Davies took over as CEO of a bank whose founder had already left. No product-market fit, no real customer base, just a banking licence and £5 million out the door in loans. Basically a blank slate with a bank charter attached.

Six years later, that bank — Allica — has lent over £3.7 billion, hit three straight profitable years, and been named the UK's fastest-growing tech company by Deloitte two years running. The interesting part isn't the growth numbers though. It's how they got there, and it's got some real lessons for anyone building product.

He'd seen the same problem from three angles

Davies didn't come up as a typical bank exec. Strategy consulting first, then almost a decade at Barclays working with SMEs. In 2013 he became the founding CEO of OakNorth, another SME-focused challenger bank. Then Revolut, where he built out their banking product across Ireland, the US, Australia, and Singapore.

So by the time he got to Allica, he'd watched the same underserved-SME problem play out inside a legacy bank, a specialist challenger, and a fast-scaling consumer fintech. That's basically 20 years of user research before writing a line of product spec.

The actual unlock: running the bank like a tech company

This is the part that'll resonate with a dev audience. Davies has said The Geek Way the book about how tech companies operate directly shaped how he restructured Allica. The core idea: strip out the layers of hierarchy and approval chains banks normally run on, replace it with the speed and ownership culture of a tech company.

They didn't outsource the tech stack either. Allica built proprietary systems in-house so relationship managers could move fast, without turning the whole thing into a faceless app. That's the real product decision here not a slicker UI, but engineering speed paired with an actual human relationship, aimed at a customer segment (established SMEs) that neither fintech apps nor legacy banks have served well.

If you've ever worked somewhere that tried to "move fast" but got strangled by approval chains, this is basically the counter-example.

AI as plumbing, not a pitch deck slide

Davies has been refreshingly unglamorous about where AI shows up at Allica. Not chatbots, not a customer-facing gimmick — it's internal. Cutting consultancy spend, speeding up product development,and streamlining the lending pipeline. The boring backend stuff that actually determines how fast a bank can say "yes" to a loan.

He's tied this to a number worth remembering: UK SMEs reportedly miss out on around £9 billion in interest every year, mostly because the products and processes built for them are outdated or just don't exist. Framed that way, the AI investment isn't innovation theater; it's closing a very specific, very quantifiable inefficiency.

The growth numbers, briefly

Quick stats for context: Fast 50 winner (fastest-growing UK company) in 2023 and 2024, Sifted's fastest-growing European startup at the end of 2024, second-fastest in Europe per the FT in 2025. ~800 employees, £371.3M revenue by 2025.

Worth noting though awards are a lagging signal, not the strategy. The actual bet stayed consistent the whole time: serve established SMEs, keep humans in the loop, skip the consumer-growth playbook. The trophies are a byproduct of sticking to that, not the plan itself.

What's next

Davies has talked about international expansion, plus pushing further into spend management and cash flow forecasting—moving from "we lend you money" toward "we're your financial operating system." The October 2025 acquisition of Kriya (embedded finance, working capital tools) is an early signal of that direction.

The actual takeaway

Strip away the banking specifics and this is a pretty clean case study in rebuilding something with no clear owner, no obvious playbook, and a founder-shaped hole in the middle. The formula: pick one underserved user, commit to one cultural bet (in this case, literally borrowing tech-company operating principles), and give it years not quarters to compound.

Whether that holds up once international expansion and new product lines start pulling the team in different directions is the part still unwritten.