- Generative Finance
- Posts
- Revolut's Roadmap + Old Fraud Data
Revolut's Roadmap + Old Fraud Data
The latest in generative AI x finance
Pranjal Daga
November 26, 2024
What’s up, everyone – Pranjal here.
It’s almost Thanksgiving. Grateful for all the readers of Generative Finance and everyone who’s supporting our journey at Accend. I hope you’re all winding down to enjoy the rest of the week with family and friends.
Let’s jump in!
My favorite finds of the week.
NEWS
Revolut’s new roadmap
The backstory: Revolut has transformed from a simple forex card into Europe's most valuable private tech company ($45B). With 50 million customers globally, they've proven challenger banks can reach traditional bank scale. But they've faced a key challenge: moving beyond their core payments product to generate meaningful revenue per customer.
Now... Revolut announced their 2025 product roadmap with four major expansions:
AI-powered financial assistant that learns user behavior
Branded ATM network starting in Spain, with facial recognition and cash deposits
Mortgage products launching in Lithuania, Ireland, and France
Business expansion including credit products and a restaurant management system (Revolut Kiosk)
THE TAKEAWAY
The average European bank makes $400 per customer annually. Revolut? Just $35. While fintech critics point to this revenue gap as proof challenger banks can't work, Revolut's 2025 strategy reveals a more nuanced truth: they're not trying to be a better bank, they're building a new category entirely.
Consider their ATM strategy - launching in Spain first isn't random. Spain has 104 ATMs per 100k adults, double the UK's 47. But unlike traditional banks who see ATMs as a cost center, Revolut's facial recognition integration suggests they see physical touchpoints as data collection opportunities. The AI assistant isn't just about personalized recommendations - it's about understanding financial behavior at a scale no traditional bank has achieved.
The bottom line: Traditional banks optimize for revenue per customer. Revolut is optimizing for data per customer. In the AI era, that might be worth more.
MY TAKE
Why your banks’s 30 year-old fraud data is… not that useful.
The "fintech has bad fraud detection" narrative is dead wrong. Traditional banks aren't better at fraud prevention because they have 30 years of data - they just have easier fraud to detect.
Banks proudly announce they have decades of transaction history while serving mostly middle-class customers with predictable spending patterns. Meanwhile, fintechs are fighting sophisticated fraud rings while serving migrants sending cross-border remittances, gig workers with irregular income, and digital natives who split bills through five different apps.
Modern financial crime makes historical data look quaint. When fraudsters are using AI to generate synthetic identities that look perfect on paper, what good is knowing how boomers spent money in the 90s? When mule networks are using thousands of real accounts to move money at lightning speed, how relevant are transaction patterns from 2005?
This is where it gets interesting. Fintechs' supposed weakness - lack of historical data - forced them to innovate in ways banks never had to:
Revolut analyzes how you hold your phone during transactions
Wise builds risk models around cross-border payment patterns that didn't exist a decade ago
Monzo spots fraud by analyzing the microseconds between how you type and swipe
The most sophisticated fraud prevention systems aren't being built by those with the most data - they're being built by those fighting the hardest fraud problems. While banks analyze historical patterns, fintechs are building AI that adapts to fraud in real-time.
This is why the "just share fraud data" solution misses the point entirely. By the time a fraud pattern shows up in historical data, criminals have already evolved. In 2024, the average sophisticated fraud attack lasts less than 24 hours. Good luck stopping that with pattern analysis from last quarter.
The future isn't about having more historical data - it's about having better real-time intelligence. As AI democratizes fraud capabilities, the only defense is AI that can spot anomalies without needing years of patterns to learn from.
Banks' 30-year data advantage is about to become as relevant as cavalry tactics in modern warfare. In the AI era, the winners won't be those with the biggest data warehouses, but those with the fastest neural networks.
The irony? Fintechs might end up teaching banks how to fight fraud, not the other way around.
Until next time, Pranjal
How I can help
We can help speed up your compliance and onboarding process.