When Brett King published Bank 2.0 in 2010, plenty of bank boardrooms dismissed as fiction his argument that banking would become something you do rather than somewhere you go. Sixteen years and several books later, that call has largely played out as he said it would. His eighth banking book, Bank 5.0: The Rise of the Agentic Bank, makes a far bigger one: the next customer of the bank may not be a human at all.
The Fintech Times has read an advance review copy ahead of the book’s global launch on 19 November 2026 at the Singapore FinTech Festival, where King delivers the Banking 5.0 keynote before a book signing with his contributing authors. Wiley publishes the global English edition and Marshall Cavendish Business the Asia edition.
The argument in one line
King states his thesis early and returns to it often: “By the middle of this century, half of all economic activity will be initiated, negotiated and executed by machines.” If he is right, the AI agents doing that work will need financial infrastructure built around them. Identity, settlement, credit decisioning and compliance were all designed for a human who signs a form, goes to sleep and waits for a reply. The back cover puts it plainly: “Those agents will need banks engineered for them. Not for us.”
Readers of the series will recognise the structure. Bank 2.0 was about self-service, Bank 3.0 about mobile and Bank 4.0 about banking everywhere. In the closing chapter King describes Bank 5.0 as something larger: “the transition from financial intermediation to economic orchestration.”
Part I: why King thinks this is inevitable
The opening five chapters make the case that AI is, in King’s words, “not overhyped; if anything, it is under-hyped”. He moves quickly from the history of automata to China’s dark factories, the labour market debate and the pace of frontier model releases, and argues that banks do not need to wait for artificial general intelligence to be disrupted. Today’s agentic tools are already enough to automate most banking workflows.
The banking experience chapter strips products back to the three utilities underneath them: storing money, moving money and accessing credit. Everything else, King argues, is a historical wrapper that generative interfaces will dissolve. “In agentic banking, the interface is the agent.”
The chapter on how fintechs “ate the banks’ lunch” is King at his most combative. He walks through each incumbent dismissal of neobanks over the past decade and uses the growth of Nubank, Revolut, WeBank and Barq to show how each one aged. On Barq, the Saudi neobank whose board he sits on, he discloses his interest. Anyone still convinced that being a solid traditional bank is enough, he warns, “will probably choke on your iced matcha latte.”
The section closes with a line that will divide bankers and regulators: “Humans will only slow down the operation of the 21st century banking industry”, King writes. A few pages later he sets out the governance model the rest of the book depends on: “humans govern objectives, boundaries and accountability; AI executes within constraints.”
A book of many voices
Bank 5.0 is not a solo effort. Nine contributing authors write or co-write chapters, and several of them are among the strongest in the book.
Dr Efi Pylarinou opens Part II by calling agentic commerce “the third rupture”, after the Silk Road and e-commerce. Her most useful observation is a blind spot she finds across the industry: “the banking industry is building infrastructure for agents as employees or co-workers. Almost nobody is building infrastructure for agents as customers.” Her account of six agent payment protocols arriving in 18 months, from Stripe and OpenAI to Visa, Mastercard, Klarna and PayPal, is a handy map of a standards race that changes month to month.
King’s own chapters on the orchestration layer and on data give the book its central technical claim: “a bank will be able to compete without a classic core banking system, but not without a data lake and foundation model strategy.” His shortlist of the main agent types (engagement, transaction, decision-making and operational) offers a practical vocabulary for any bank drawing up an architecture.
André Vellozo’s chapter on data savings is the most original in the book. Today’s agent payment stack only looks outward, treating the human as a source of funds and authority. Vellozo turns that round: the identity binding that makes an agent accountable can also make its owner a beneficiary. “A binding strong enough to assign blame is, by construction, strong enough to assign benefit.” Agent output settles into an owner-side “dWallet”, which he calls “the largest unclaimed deposit base in the history of banking.”
David G.W. Birch then makes the case for Know Your Agent with his usual wit and directness: “Without KYA, we can’t have nice things.” His conclusion that banks should provide the trust infrastructure for agentic commerce, because “AI is not by itself a defense against AI, but real cryptography is”, is the clearest strategic opening for incumbents in the book.
The tech stack, from Huawei’s vantage point
Part III is written by Jason Cao, CEO of Huawei Digital Finance, and edited by King. It is the most engineering-led section. It covers a resilience standard Cao calls the “Four Zeros” (zero downtime, zero wait, zero trust and zero touch), cloud-native cores and a real-time data layer. Cao’s distinction between the AI a customer sees and the capability the bank builds behind it is sharp: “Customer AI is the expression; Bank AI is the capability. One cannot outperform the other.” So is his argument that behavioural data, not the model, becomes the moat: “Google trains their engines on search. TikTok trains theirs on recommendations. Amazon trains on shopping behavior. What will banks train on?”
Readers should know that this is a vendor’s perspective. The headline metrics are not independently sourced, and the Chinese core migration case studies do not name the banks involved.
The half-second market
Part IV is where the book is most vivid. J.K. Khalil, formerly Division President at Mastercard, writes three chapters that turn the agentic bank from architecture into lived experience. Products give way to intent: “A mortgage is not the desire. A home is.”
His “junk drawer” passage is a good one. It inventories the value most of us leave scattered across expiring air miles, prepaid cards and forgotten wallets, and sets it against loyalty economics: “Breakage was the silent profit line.” Once agents route spending to wherever value is greatest, that line disappears. The same goes for pricing. The half-second between a customer’s yes and the confirmation screen becomes a live negotiation between customer, merchant, bank, lender and insurer agents. His message to incumbents is reassuring and pointed at once: “agents do not punish margin. They punish opacity.”
Khalil does not duck the risks. Ambient, in-context credit, he warns, “done poorly”, becomes “the most efficient predatory lending machine ever built”. On accountability he is blunt: “A bank cannot say, ‘The AI did it’.”
King’s chapter on financial inclusion raises what he calls the inclusion paradox: “AI may give every citizen access to world-class financial services while simultaneously reducing the number of citizens earning sufficient income to use them.” Jonathan Holman, CEO of Nebras Open Finance, the UAE’s national open finance operator, closes the section with a measured look at what agents built on open banking rails could mean for the bank’s ownership of the customer: “The teller fronting all financial services may return, but it might not be employed by the bank at all.”
Regulation in the rail, and some wider territory
Part V widens the lens to economics, regulation, health and organisational design. King’s regulation chapter argues that supervision has to move “into the rail”, with rules written as executable code and agents supervising agents. Some of his lines are made for conference slides (“The EU is the world’s best rule-writer and a mediocre rule-executor”). Singapore, Brazil and the UAE come out ahead of the US and Australia, because rulemaking, supervision and payments infrastructure sit closer together in those markets.
The economics chapter, co-written with Professor Richard Petty, argues that energy and grid capacity, not chips, will be the binding constraint on the agentic economy. It introduces a new readiness index that ranks China first, the UAE second and the US fifth. The authors say openly that the scores are “directional” and were built with help from AI models, and that the index is designed to be argued with. Dr Julia Michelin asks whether health is becoming a form of capital. Ross Dawson’s Human+AI framework is the most practical guidance in this part on keeping human judgment meaningful: “An approver who rubber-stamps a recommendation has added a signature, not a judgment.”
The roadmap
For bank leaders, the conclusion is the most useful part of the book. King argues that being a capable digital bank is now table stakes, and also a trap, because such a bank is “still fundamentally architected around human latency”. He puts the window to move from digital-first to agentic at two to three years: “Institutions whose AI agents are self-improving their own operations by 2029 will be uncatchable by 2035.”
He follows with a self-assessment roadmap built on five pillars: leadership and culture, technology foundations, organisational evolution, autonomous infrastructure and the agentic bank itself. A three-year transformation plan then runs across culture, technology, organisational structure and AI capability. There is a welcome dose of realism: King warns against “scorecard theater (self-scored 4s, when most banks are 1 or 2 genuinely)”.
His picture of 2050 ranks the winners as AI-native banks born after 2028, then fintechs that made the leap early, with adapted incumbents bringing up the rear, which he expects to be “the smallest representative group”.
Where readers will push back
Bank 5.0 is a deliberately provocative book, and not every reader will accept its certainty. King writes that “the destination is certain, but the journey is less so”, and some of his timelines are bold. The geography leans towards China, Singapore and the Gulf, where King now lives and advises government and industry, though he is open about those roles. Part III carries a vendor’s perspective. The later chapters stray into energy policy, climate, universal basic income and longevity, which will strike some readers as a long way from the bank balance sheet. And because the book captures a field moving month by month, some of its 2026 references will date quickly.
Chris Skinner, one of the book’s endorsers, puts this best: “You may not agree with every forecast, but that is precisely why you should read this book.”
The verdict
King has a long record of being early, and Bank 5.0 is his most ambitious book yet. It is also more of an ensemble piece than its predecessors, and the contributed chapters from Vellozo, Birch, Khalil and Holman give it depth beyond the author’s own forecasts. Bank CEOs, boards and strategy teams will get the most from it, and the closing roadmap alone justifies the time. Fintech founders building for agentic commerce will find a useful map of where incumbents are weakest.
In keeping with the subject, King is transparent about method. He used AI tools for research, planning and restyling the charts, but says that “all of the text that appears is human-generated.”
His closing line sums up the ambition of the whole series: “Banking everywhere, never with a human. Welcome to Bank 5.0.”
Bank 5.0: The Rise of the Agentic Bank launches on 19 November 2026 at the Singapore FinTech Festival. Details and pre-orders are at bank5book.com.
The post Brett King’s Bank 5.0 And The Rise Of The Agentic Bank appeared first on The Fintech Times.
