Skip to content

Inside Revolut’s AI Banking Brain

Read the Portuguese translation →

Author: Bernard Marr · Watch on YouTube

Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.

Original language: English

Original video: Inside Revolut’s AI Banking Brain · Channel: Bernard Marr · 01:02 · Watch on YouTube

Topics: AI agents, Enterprise AI applications, Task and workflow automation, AI models

Revolut is developing Pragma, a proprietary AI foundation model designed to understand financial behavior. Instead of using separate AI systems for different banking functions, the company connects them through a shared model trained on 40 billion interactions from 25 million users. Revolut reports improvements in fraud detection, credit risk prediction, and product recommendations, while its AI assistant handles most support requests. The video argues that businesses can gain an advantage from AI that learns across the entire organization.

  • Pragma is Revolut’s proprietary AI foundation model, trained specifically to understand financial behavior.
  • The model supports tasks including fraud detection, risk prediction, product recommendations, and money management.
  • Revolut connects banking functions through one shared model rather than relying on separate AI systems.
  • Pragma was trained on 40 billion interactions from 25 million users.
  • Revolut reports a 64.7% improvement in fraud detection.
  • The company reports a 16% increase in credit risk prediction and 41% better product recommendations.
  • Revolut’s AI assistant handles 75% of support requests.
  • Its new agent, Air, can manage subscriptions, cancel lost cards, help with budgeting, and arrange travel.
  • The video presents organization-wide learning as a potential source of competitive advantage.
Start Chapter Summary
00:00 Revolut’s AI banking model The video introduces Pragma, a proprietary model designed to support several banking tasks by understanding financial behavior.
00:30 A shared model and reported results Pragma’s training scale and reported improvements in fraud detection, credit risk prediction, and product recommendations are described.
01:02 AI assistance and the wider lesson The video outlines what Revolut’s assistant and Air can do, then argues that AI learning across an organization can create competitive advantage.

It’s called Pragma, a proprietary AI foundation model trained specifically to understand financial behavior. — 00:00

shared model trained on 40 billion interactions from 25 million users. — 00:30

Revolut reports a 64.7% improvement in fraud detection — 00:30

AI assistant already handles 75% of support requests. — 01:02

Competitive advantage will come from AI that learns across the entire organization — 01:02

Revolut, Pragma, Air

00:00 Revolut is building an AI brain for banking, one that can spot fraud, predict risks, recommend products, and help manage your money. It’s called Pragma, a proprietary AI foundation model trained specifically to understand financial behavior. Most banks use separate AI systems for fraud, credit, customer service, and marketing. Revolut connects everything through one

00:30 shared model trained on 40 billion interactions from 25 million users. Think of it like replacing a collection of specialists who rarely speak with one highly informed team that learns together. The results are impressive. Revolut reports a 64.7% improvement in fraud detection, a 16% increase in credit risk prediction, and 41% better product recommendations. Its

01:02 AI assistant already handles 75% of support requests. While its new agent, Air, can manage subscriptions, cancel lost cards, help with budgeting, and even arrange travel. The bigger lesson for every business is clear. Competitive advantage will come from AI that learns across the entire organization, allowing intelligence to grow with every interaction.