Fintechs and neobanks describe their AI work in IPO prospectuses, on earnings calls and in product launches, and some have reported its effect on headcount and customer service in detail — but rarely in one place. These pages assemble the public record for each company: the assistant or platform it runs, where AI is in production, the numbers it has disclosed, who leads the work, which licences and regulators it answers to, and what the record suggests. Every dated claim links to its source. The same format covers the 100 largest US banks on the bank pages.
NYSE: NU · Brazil
Nubank (Nu Holdings)
Build in-house: a proprietary transaction foundation model (nuFormer) at the centre of underwriting, support and growth decisions, stated as an 'AI-first' vision since Q3 2025, with the CEO now on OpenAI's boards.
Private · United Kingdom
Revolut
Build, don't bolt on: an in-house research division trains a transaction foundation model that is meant to feed the assistant, fraud and credit systems from one shared engine.
NASDAQ: CHYM · United States
Chime
AI on a proprietary core: Chime runs AI through ChimeCore, its own processor and ledger, first to automate support and engineering, then to turn Jade into a co-pilot that acts on members' money with permission.
Private · United Kingdom
Monzo
Build in-house, vendor-neutral: Monzo builds its own LLM agents for support, fraud and engineering on an internal AI gateway that connects to multiple model vendors, pairing LLMs with supervised ML models and human review.
Private · United Kingdom
Starling Bank
Gemini inside a proprietary stack: Starling builds customer-facing AI tools on its own banking platform using Google Gemini on Google Cloud, rolling them into one agentic assistant, while cutting roles to fund more AI.
NASDAQ: AFRM · United States
Affirm
In-house machine learning that underwrites every purchase in real time, now on a transformer model, plus an engineering organisation retooled around coding agents and a push to put pay-over-time inside AI shopping agents.
NYSE: KLAR · Sweden
Klarna
AI first, then a correction: an OpenAI-built assistant and a hiring freeze shrank headcount by about 40%, human agents came back, and AI now faces outward into agentic shopping.
NASDAQ: HAPN · United States
Happen Bank (formerly LendingClub)
A marketplace lender turned bank that treats AI as an operating-efficiency lever: machine-learning underwriting at the core, then AI automation of loan operations and servicing, including an AI member-service agent called Penny.
NASDAQ: UPST · United States
Upstart
An AI underwriting company from the start: in-house credit models sold to banks and credit unions as a marketplace, now seeking its own national bank charter built around those models.
NASDAQ: PYPL · United States
PayPal
A payments incumbent in turnaround that uses long-standing AI for fraud and risk, is betting on agentic commerce across ChatGPT, Copilot, Perplexity and Gemini, and under a new CEO is using AI inside the company to cut costs.
NYSE: XYZ · United States
Block (Square, Cash App)
AI as the operating model: Block halved its workforce in February 2026 to run with smaller AI-using teams, builds its own model-agnostic agent stack (goose, Builderbot), and ships proactive agents to consumers (Moneybot) and sellers (Managerbot).
Private · United States
Stripe
AI as both engine and market: machine-learning models (now a payments foundation model) run fraud and authorization inside the network, while Stripe sells 'economic infrastructure for AI' to AI companies and to the agents that buy on consumers' behalf.
Euronext Amsterdam: ADYEN · Netherlands
Adyen
AI built into the payment rails: a single platform and in-house data feed Uplift's optimisation models and the Dynamic Identification trust layer, which Adyen now extends outward to AI shopping agents through Adyen Agentic.
NASDAQ: WSE · United Kingdom
Wise
AI behind the counter, under compliance pressure: machine learning runs compliance checks and fraud alerts and LLM copilots support financial-crime and support staff, while regulators in Belgium and the US question its AML controls.
NYSE: V · United States
Visa
Network-as-infrastructure for AI agents: Visa uses AI to defend the network and sells AI fraud tools to banks, while opening its credentials, tokens and rails to agent builders through Visa Intelligent Commerce.
NYSE: MA · United States
Mastercard
Trust layer for agent payments plus AI-powered services: Mastercard registers and tokenizes AI agents (Agent Pay), proves what a user authorized (Verifiable Intent), and sells AI fraud, cyber and advisory services built on its network data.
NASDAQ: HOOD · United States
Robinhood
AI as the interface to the brokerage: Robinhood Cortex answers and acts inside the app for Gold subscribers, while Agentic Trading opens ring-fenced accounts to customers' own third-party AI agents through Robinhood's MCP servers.
Private · United States
Ramp
Agents across the finance back office, plus a new line in managing AI spending itself: Ramp sells agents for expense review, bills, procurement and accounting, and since April 2026 tools to track and route companies' AI token spend.
NYSE: COF · United States
Brex
Agentic spend management inside a bank: Brex sells AI agents that collect receipts, review, audit and book corporate expenses, and since April 2026 it does so as a Capital One subsidiary that its founder still runs.
Private · United States
Plaid
Data network turned AI layer: Plaid trains its own foundation models on transaction data from its account-connection network and sells the connections that let AI assistants (ChatGPT, Meta's Muse, Sierra agents) read users' bank accounts.
Which fintechs and neobanks are covered?
20 so far: neobank — Nubank, Revolut, Chime, Monzo and Starling; bnpl & lending — Affirm, Klarna, Happen Bank and Upstart; payments — PayPal, Block, Stripe, Adyen and Wise; card network — Visa and Mastercard; brokerage — Robinhood; business finance & infrastructure — Ramp, Brex and Plaid. More companies are being added; a page is published only once every dated claim on it is sourced.
How is this section sourced?
Every dated statement on a company page cites a numbered source: the company's own newsroom, investor-relations filings and annual reports, regulator registers and statements, or reporting by Reuters, the Financial Times, CNBC, Bloomberg, the Wall Street Journal and the specialist fintech press. 320 sources across the 20 pages; blogs and aggregators are excluded, and numbers are quoted as reported with their date.
How do the fintech pages differ from the bank pages?
They use the same structure — answer first, a dated timeline, a use-case table with status, disclosed numbers, quotes, named leaders, regulators, analysis and numbered sources — so a fintech and a bank can be compared line by line. The facts at the top differ: instead of Federal Reserve rank and lead-bank assets, a fintech page gives its category, founding year, licences and supervisors, and one sourced scale figure. The 100 bank pages are at /banks.
Every fintech's next AI move, the morning after.
when a bank, fintech or regulator moves on AI, it is in the next morning's brief · six sourced stories · 7 am ET · free
plus every tracker, bank and agent page update, the morning after · leave any morning